reproductions supplied by edrs are the best that can be ... · pdf filetable. 1.1. 2.1....
Post on 08-Mar-2018
214 Views
Preview:
TRANSCRIPT
DOCUMENT RESUME
ED 452 292 UD 034 113
AUTHOR Morris, Pamela A.; Huston, Aletha C.; Duncan, Greg J.;Crosby, Danielle A.; Bos, Johannes M.
TITLE How Welfare and Work Policies Affect Children: A Synthesisof Research.
INSTITUTION Manpower Demonstration Research Corp., New York, NY.SPONS AGENCY David and Lucile Packard Foundation, Los Altos, CA.; John D.
and Catherine T. MacArthur Foundation, Chicago, IL.; Grant(W.T.) Foundation, New York, NY.; ARCO Foundation, LosAngeles, CA.; Ford Foundation, New York, NY.; Ewing MarionKauffman Foundation, Kansas City, MO.; Alcoa Foundation,Pittsburgh, PA.; George Gund Foundation, Cleveland, OH.;Anheuser-Busch Companies, Inc., St. Louis, MO.; Open SocietyInst., New York, NY.; New York Times Foundation, NY.
PUB DATE 2001-03-00NOTE 126p.; Funding also provided by the Ambrose Monell
Foundation, Heinz Family Foundation, the Grable Foundation,and the Union Carbide Foundation. Special funding wasprovided by BP on behalf of ARCO to support the publicationand dissemination of this document. Part of the NextGeneration Project.
AVAILABLE FROM Manpower Demonstration Research Corporation, 16 East 34Street, New York, NY 10016. Tel: 212-532-3200; Web site:http://www.mdrc.org.
PUB TYPE Reports Evaluative (142)EDRS PRICE MF01/PC06 Plus Postage.DESCRIPTORS Academic Achievement; Child Health; Elementary Education;
Employment; *Family Income; Interpersonal Competence; *OneParent Family; Poverty; Program Effectiveness; *WelfareRecipients; *Welfare Reform; *Welfare Services
IDENTIFIERS Time Limited Welfare
ABSTRACTThis monograph synthesizes findings from five large-scale
studies investigating welfare and employment programs in order to examine theeffects on children of key policies: providing financial support for workingfamilies, requiring single parents to work, and limiting time on welfare. Itidentifies program features associated with effects on children's schoolachievement, social behavior, and health. The studies randomly placed parentsin either a program group, which had access to new services and benefits andwas subject to new rules, or a control group, which received benefits and wassubject to previously existing rules. In surveys conducted 2-4 years afterparents entered the studies, children's school achievement, social behavior,and health were measured, and a subset of single parents was analyzed.Overall, programs that included earnings supplements for parents who went towork positively affected elementary students, whose academic achievementsurpassed that of children in the traditional welfare system. Addingmandatory employment services did not greatly reduce those positive effects.Programs with mandatory employment services and programs with time limits hadfew effects on children. They did have some negative effects on adolescentbehavior and school achievement. The positive effects of earnings supplementprograms on children were most pronounced for children of long-term welfare
Reproductions supplied by EDRS are the best that can be madefrom the original document.
recipients. An appendix presents outcome levels and impacts underlying effectsizes presented in this monograph. (Contains 76 references.) (SM)
Reproductions supplied by EDRS are the best that can be madefrom the original document.
I- -
i 4! .i - Alk
I . ini Al
. Ait 0.
i5
10 .._
I/. A
LA :
U S DEPARTMENT OF EDUCATIONOffice of Educational Research and Improvement
EDUCATIONAL RESOURCES INFORMATIONCENTER (ERIC)
III(This document has been reproduced asreceived from the person or organizationoriginating it
Minor changes have been made toimprove reproduction quality
Points of view or opinions stated in thisdocument do not necessarily representofficial OERI position or policy
PERMISSION TO REPRODUCE ANDDISSEMINATE THIS MATERIAL HAS
BEEN GRANTED BY
1".S.G--vein a iltikan DerrA o e_5_es
1
TO THE ED CATIONAL RESOURCESINFORMATION CENTER (ERIC)
2I II Mk
AI
. 01
tiD
BOARD OF DIRECTORS
PAUL H. O'NEILL, ChairmanChairmanAlcoa
ROBERT SOLOW, Vice ChairmanInstitute ProfessorMassachusetts Institute of Technology
RUDOLPH G. PENNER, TreasurerSenior FellowUrban Institute
MARY JO BANEProfessor of Public PolicyJohn E Kennedy School of GovernmentHarvard University
REBECCA M. BLANKDeanGerald R. Ford School of Public PolicyUniversity of Michigan
JAMES H. JOHNSON, JR.E. Maynard Adams Professor of Business,
Geography, and SociologyDirector, Urban Investment Strategies CenterUniversity of North Carolina
RICHARD J. MURNANEProfessor of EducationGraduate School of EducationHarvard University
MARION 0. SANDLERChairman and CEOGolden West Financial Corporation and
World Savings and Loan Association
ISABEL V. SAWHILLSenior FellowBrookings Institution
LAWRENCE J. STUPSKIChairmanStupski Family Foundation
WILLIAM JULIUS WILSONMalcolm Wiener Professor of Social PolicyJohn E Kennedy School of GovernmentHarvard University
JUDITH M. GUERONPresidentManpower Demonstration Research Corporation
MDRC
TheII.diext . nneration
How Welfare and Work Policies
Affect Children:
A Synthesis of Research
Pamela A. MorrisAletha C. HustonGreg J. DuncanDanielle A. CrosbyJohannes M. Bos
MDRC
Manpower. Demonstration Research Corporation
March 2001
The Next Generation Project
This report is part of the Next Generation, a project that examines the effects of welfare,antipoverty, and employment policies on children and families. Drawing on rich datafrom recent welfare reform evaluations, the project aims to inform the work ofpolicymakers, practitioners, and researchers by identifying policy-relevant lessons thatcut across evaluations.
Foundation partnersThe Next Generation project is funded by the David and Lucile Packard Foundation,William T. Grant Foundation, and John D. and Catherine T. MacArthur Foundation.
Research partnersThe project is a collaboration among researchers from MDRC, the University of Texas atAustin, Northwestern University, the University of California at Los Angeles, Kent StateUniversity, the University of Michigan, New York University, and the Social Researchand Demonstration Corporation.
Project directorVirginia Knox, Senior Research Associate, MDRC, 16 East 34 St., New York, NY 10016E-mail: virginia_knox@mdrc.org; phone: (212) 340-8678
www.mdrc.org/NextGeneration
Special funding was provided by BP on behalf of ARCO to support the publication anddissemination of this document.
Dissemination of MDRC publications is also supported by the following foundations andindividuals who help finance MDRC's public policy outreach and expanding efforts tocommunicate the results and implications of our work to policymakers, practitioners, and others:the Ford, Ewing Marion Kauffman, Ambrose Monell, Alcoa, George Gund, Grable, Anheuser-Busch, New York Times Company, Heinz Family, and Union Carbide Foundations; and the OpenSociety Institute.
For information about MDRC and copies of our publications, see our Web site: www.mdrc.org.MDRC® is a registered trademark of the Manpower Demonstration Research Corporation.
(
Copyright © 2001 by the Manpower Demonstration Research Corporation. All rights reserved.
Contents
List of Tables, Figures, and BoxesPreface viiAcknowledgments viiiExecutive Summary ES-1
1. Introduction 1
I. Historical Background: Welfare Reform and Children 2II. Program Features 2
A. Earnings Supplements 3
B. Mandatory Employment Services 4C. Time Limits 4
DI The Relation Between Program Features and Children's Outcomes 4A. Effects of Maternal Employment 6B. Effects of Family Income 6C. Effects of Welfare Receipt 7D. Differences by Age 7E. Differences by Gender 8
IV. Examining Children in Random Assignment Studies 9V. The Studies 9
A. The Minnesota Family Investment Program (MFIP) 10B. The Self-Sufficiency Project (SSP) 11
C. The New Hope Program 11
D. The National Evaluation of Welfare-to-Work Strategies (NEWWS) 12E. Florida's Family Transition Program (FTP) 12
VI. Comparisons Across Programs 13VII. Time Periods of the Studies 14VIII. Sample, Measures, and Analysis Strategy 15
A. Sample 15B. Measures 16C. Analysis Strategy 17
2. Effects on Children of Programs with Earnings Supplements 18
I. Effects on Parents' Economic Outcomes 18II. Effects on Children 19III. Adding Mandatory Employment Services to an Earnings Supplement 28IV. How Might These Programs Have Affected Children? Effects on
Child Care, Parents' Emotional Well-Being, and Parenting Behavior 29V. How Did Children and Families in the Program Groups Fare? 31VI. Effects on Long-Term Welfare Recipients 33VII. Differences Between Preschool-Aged and Early School-Aged Children 35VIII. Differences Between Boys and Girls 35Dc. Summary and Discussion of the Effects of Programs with
Earnings Supplements 39
6
3. Effects on Children of Programs with Mandatory EmploymentServices 43
I. Effects on Parents' Economic Outcomes 43II. Effects on Children 44III. Effects on Child Care, Parents' Emotional Well-Being, and Parenting
Behavior 50IV. Summary and Discussion of the Effects of Programs with Mandatory
Employment Services 50
4. Effects on Children of Programs with Time Limits 52
I. Effects on Parents' Economic Outcomes 52II. Effects on Children 53III. Summary and Discussion of the Effects of Programs with Time Limits 55
5. Effects of Welfare and Employment Programs on Very YoungChildren and Adolescents 57
6. Conclusions and Policy Implications 61
I. Policy Implications 63II. Further Research from the Next Generation Project 64
Appendix: Outcome Levels and Impacts Underlying the Effect SizesPresented in This Monograph 67
ReferencesRecent Publications on MDRC Projects
-iv-
7
O
8995
Table
1.1
2.1
AppendixTable
List of Tables, Figures, and Boxes
Features of Each Program, by Study
Program Group Outcomes for Earnings Supplement Programs
Page
14
32
1 Earnings Supplement Programs: Impacts for Children Who Were Preschool-Aged or Early Elementary School-Aged at Random Assignment 72
2 Earnings Supplement Programs: Impacts for Children of Long-Term RecipientsWho Were Preschool-Aged or Early Elementary School-Aged at RandomAssignment 74
3 Earnings Supplement Programs: Impacts for Children Who Were Preschool-Aged or Early Elementary School-Aged at Random Assignment, by Children'sAge 76
4 Earnings Supplement Programs: Impacts for Children Who Were Preschool-Aged or Early Elementary School-Aged at Random Assignment, by Children'sGender 79
5 Programs with Mandatory Employment Services: Impacts for Children WhoWere Preschool-Aged at Random Assignment 82
6 Programs with Mandatory Employment Services: Impacts for Children of Long-Term Recipients Who Were Preschool-Aged at Random Assignment 84
7 Time-Limited Program (FTP): Impacts for Children Who Were Preschool-Agedor Elementary School-Aged at Random Assignment 86
8 Two Different Programs (SSP and FTP): Impacts for Adolescent Children 87
Figure
1.1 Mechanisms Through Which Welfare and Employment Policies Might AffectChildren
2.1 The Three Earnings Supplement Programs Without Mandatory EmploymentServices Improved Children's School Achievement
5
22
2.2 One Earnings Supplement Program Without Mandatory Employment ServicesReduced Children's Behavior Problems 23
2.3 Two Earnings Supplement PrograMs Increased Children's Positive Behavior 26
2.4 One Earnings Supplement Program Improved Children's Health as Reported byParents 27
2.5 Adding Mandatory Employment Services to an Earnings Supplement ProgramAffected None of the Effects for Children Except Positive Behavior 30
-v-
Figure Page
2.6 The Effects on School Achievement Did Not Differ Consistently AcrossEarnings Supplement Programs by Children's Age at Random Assignment 36
2.7 Earnings Supplement Programs Decreased Behavior Problems Somewhat Morefor Children Who Were Older Than for Children Who Were Younger atRandom Assignment 37
2.8 The Effects on School Achievement Did Not Differ for Boys and GirlsConsistently Across Earnings Supplement Programs 38
2.9 The Effects on Behavior Problems Did Not Differ for Boys and GirlsConsistently Across Earnings Supplement Programs 40
3.1 For Children Aged 3-5 at Random Assignment, Programs with MandatoryEmployment Services Had Few Effects on School Achievement 45
3.2 For Children Aged 3-5 at Random Assignment, Programs with MandatoryEmployment Services Did Not Consistently Reduce or Increase BehaviorProblems 47
3.3 For Children Aged 3-5 at Random Assignment, Two Programs with MandatoryEmployment Services Had Negative Effects on Health 49
4.1 The Only Study of Time-Limited Welfare Found Few and Mixed Effects onChildren's Outcomes 54
5.1 SSP Increased Adolescents' Behavior Problems 58
5.2 FTP Had Two Unfavorable Effects on Adolescents 59
6.1 Summary of All 11 Programs' Impacts on Children's School Achievement 62
Box
1.1 Studies Examined in This Monograph 10
2.1 How to Read the Figures in This Monograph 21
2.2 How Large Are These Effect Sizes? 24
9-vi-
Preface
This is the first report from the Next Generation project, an innovative collaborationamong researchers at MDRC, several other leading research institutions, and the foundationfunding partners that is aimed at understanding the effects of welfare and employment policieson low-income children and families. The collaborative and interdisciplinary nature of the pro-ject is reflected in the combination of authors of this document Pamela Morris and JohannesBos at MDRC, Aletha Huston and Danielle Crosby at the University of Texas at Austin, andGreg Duncan at Northwestern University who together represent the fields of developmentalpsychology, economics, and policy analysis.
The monograph provides the first comprehensive look at the findings from several recentevaluations of welfare and employment programs in order to examine the effects on children ofthree key policy approaches: providing financial supports to working families, requiring singleparents to work, and limiting the length of time families can receive welfare. The studies onwhich this work is based were begun prior to the landmark federal welfare reforms of 1996, butmany states have incorporated one or more of these policies into their post-1996 programs.
The most consistent finding is that programs that provided financial supports to parentswho went to work and increased parental employment and family income as a result im-proved outcomes for children. Four of the 11 programs examined here offered such financialsupports; in all four, elementary school-aged children's school achievement was higher than thatof children whose families were in the traditional welfare system. Thus, it appears that such pro-grams have the potential not only to support the working poor but also to complement educationreforms aimed at improving the school achievement of low-income children.
The document also provides some reassurance about the effects on children of requiringsingle parents to participate in work-related activities. The six programs examined here that in-creased parental employment through such mandatory employment services showed little evi-dence of negatively affecting elementary school-aged children, and they saved the governmentmoney. However, these programs also showed little evidence of benefiting these children. Re-garding older children, for whom outcomes were examined in two of the studies included here,the report sounds a note of caution: Both programs increased parental employment but had somenegative effects on adolescents' behavior and school achievement.
Overall, the findings suggest that policymakers face a choice between offering mandatoryemployment services without financial work supports, which increase parental employment andreduce welfare dependence but have only neutral effects on children, and providing financialwork supports, which increase parental employment, boost family income, and benefit childrenbut also raise government expenditures.
This monograph represents the kind of cross-cutting research synthesis one directlyrelevant to policymakers that is the mission of the Next Generation project. The project's con-tinuing work will provide more detailbd analyses of how job characteristics, child care policies,and family income affect low-income children.
10-vii-
Judith M. GueronPresident
Acknowledgments
We are indebted to the Next Generation project's original director, Robert Granger, forhis vision and dedication to developing the project. We are also grateful to the project's funders
the David and Lucile Packard Foundation, William T. Grant Foundation, and John D. andCatherine T. MacArthur Foundation for their support, enthusiasm, and recognition of thevalue of this multistudy, collaborative effort.
The Next Generation project would not be possible without the program evaluations onwhich it is based, all of whose sponsors generously permitted us to use the original data. Thoseevaluations were supported by numerous federal agencies, state agencies, and foundations, whichare thanked individually in the evaluation reports.
The Next Generation project has benefited in particular from the foresight and leadershipshown by officials at the U.S. Department of Health and Human Services (HHS), Office of theAssistant Secretary for Planning and Evaluation (ASPE) and Administration for Children andFamilies (ACF), in their role in the original evaluations. ASPE helped to pioneer research onwelfare and employment programs' effects on children by including information on children'soutcomes in the National Evaluation of Welfare-to-Work Strategies (NEWWS). ACF developedthe Project on State-Level Child Outcomes, an effort currently under way to include comparablemeasures of children's well-being in five state welfare reform evaluations.
Several of the evaluations benefited from the work of researchers at Child Trends, par-ticularly that of Kristin Moore, Martha Zaslow, and Sharon McGroder. Child Trends is conduct-ing the Child Outcomes Study in the NEWWS evaluation under subcontract to MDRC. ChildTrends also played a central coordinating role in the Project on State-Level Child Outcomes,which is a collaboration among Child Trends, HHS (as initiator and funder), MDRC (as evalua-tor in Minnesota, Florida, and Connecticut), Mathematica Policy Research (as evaluator in Iowa),and Abt Associates (as evaluator in Indiana), with input from state participants and the NationalInstitute of Child Health and Human Development (NICHD) Family and Child Well-Being Re-search Network. Child Trends was particularly instrumental in the conceptualization and designof the survey modules focused on children's outcomes.
A number of reviewers provided invaluable feedback on this monograph. Gayle Hamiltonand Lisa Gennetian at MDRC provided key guidance throughout the process. In addition, Vir-ginia Knox, Judith Gueron, Judith Greissman, Barbara Goldman, and Gordon Berlin at MDRCoffered critical insights into the analysis and presentation of the findings. The report also bene-fited from input and comments from Kristin Moore, Martha Zaslow, and Sharon McGroder atChild Trends; Martha Moorehouse, Audrey Mirsky-Ashby, and Howard Rolston at HHS; andHiro Yoshikawa at New York University. We would also like to thank Christopher Jencks for hisfeedback on an early presentation of the findings.
11-viii-
The monograph benefited tremendously from the dedication of Phuong Tang, who cre-ated all the tables and figures, fact-checked the report, and coordinated production of the docu-ment. Thanks are also due Frank Tsai and Wanda Vargas at MDRC for conducting analyses forthe monograph and Young Chang at the University of Texas for her assistance in the analysis andpresentation of the results. Christina Gibson and Katherine Magnusson at Northwestern Univer-sity also provided invaluable assistance. The monograph was edited by Valerie M. Chase with
the assistance of Robert Weber, and Stephanie Cowell did the word processing.
The Authors
Executive Summary
Over the past 30 years, welfare and other public policies for families living in povertyhave developed a primary objective of increasing parents' self-sufficiency by requiring and sup-porting employment. The Personal Responsibility and Work Opportunity Reconciliation Act(PRWORA), passed in 1996, was a milestone in this effort, limiting the length of time that fami-lies can receive federal cash welfare assistance and requiring most of them to participate in em-ployment-related activities to be eligible for such assistance. In addition, during the 1990s themaximum benefits available to working-poor families through the Earned Income Credit (thefederal tax credit that supplements the earnings of low-income families), publicly funded healthinsurance, and child care assistance were expanded to reward work outside the welfare system.Because many of these benefit expansions encourage parental employment, and because otherchanges have weakened the safety net for families in which parents do not maintain employment,all these developments may have important consequences for children.
Proponents of changes in welfare policy have argued that parental employment benefitschildren by providing them with family role models who work and are self-sufficient and by in-troducing a regular schedule into the family routine. But employment may also create stress inthe family, reduce parents' opportunities to spend time with their children, and interfere withparents' monitoring of their children's activities particularly in single-parent families. Chil-dren may also be influenced by parental employment through changes in family resources: Iffamily income or subsidies supporting such work-related needs as child care increase, childrenmay benefit; if family resources decrease, children may be harmed. The critical question for pol-icy is not "What are the effects of welfare reform on children?" Instead, it is "What program fea-tures are most likely to promote children's well-being?" or, conversely, "What' program featuresharm children or leave them unaffected?"
In this monograph, we synthesize the results of five large-scale studies (see text box) thattogether examine the effects on children of 11 different employment-based welfare and antipov-erty programs aimed primarily at single-parent families. (A companion document' examines theeffects of these and other programs on parental employment, welfare use, and income.) Specifi-cally, we attempt to identify the program features that are associated with effects on children'sschool achievement, social behavior, and health. Although all the studies were under way by1996, they were designed to test the effects of many program features that have been imple-mented by the states since the federal welfare law of 1996 was passed. The monograph is a prod-uct of the Next Generation project, a collaboration among researchers at the ManpowerDemon-stration Research Corporation (MDRC) and several leading research institutions that is beingfunded by the David and Lucile Packard Foundation, William T. Grant Foundation, and John D.and Catherine T. MacArthur Foundation.
We classify the programs in these studies on the basis of three features that might have af-fected the experiences of children in the participating families:
How Welfare and Work Policies Affect Employment and Income: A Synthesis of Research (MDRC). 2001. DanBloom and Charles Michalopoulos.
ES-1
Studies Examined in This Monograph
The Next Generation project analyzes data from five program evaluations, building on their re-search designs, outcome measures, and impact analyses. The evaluations, and the organizationsthat conducted them, are listed below.
Florida's Family Transition Program was evaluated by MDRC under contract to the Florida De-partment of Children and Families.
The Minnesota Family Investment Program was evaluated by MDRC under contract to the Min-nesota Department of Human Services.
The National Evaluation of Welfare-to-Work Strategies is being conducted by MDRC under con-tract to the U.S. Department of Health and Human Services. The Child Outcomes Study, whichexamines program impacts on young children, is being conducted by Child Trends under subcon-tract to MDRC.
The New Hope program is being evaluated by MDRC under contract to the New Hope Project,Inc., in collaboration with researchers from Northwestern University, the University of Texas atAustin, the University of Michigan, and the University of California at Los Angeles.
The Self-Sufficiency Project was conceived by Human Resources Development Canada. The pro-ject is being managed by the Social Research and Demonstration Corporation (SRDC) and evalu-ated by SRDC and MDRC.
1. Earnings supplements. Four of the programs offered generous earnings sup-plements designed to make work more financially rewarding by providingfamilies with cash supplements or by increasing the amount of welfare thatparents could keep when they went to work. (One of the programs also sup-plemented earnings less directly by subsidizing child care and health carebeyond the levels provided in the community.) Earnings supplements are in-tended to increase family resources as well as to encourage parental employ-ment, and in the programs under study they generally succeeded in achievingboth of these goals. While some of the programs with earnings supplementsincluded other components as well, the provision of supplements was the onlyfeature that the four programs in this category shared.
2. Mandatory employment services. Six of the programs provided only manda-tory employment services such as education, training, or immediate jobsearch in which parents were required to participate to be eligible to receivecash welfare benefits. Parents who failed to comply were subject to sanctionsin the form of reduced welfare grants. The six programs in this category in-cluded mandatory employment services without any earnings supplements ortime limits. In the programs under study, participation mandates (designedprimarily to increase employment) were generally successful in raising em-ployment rates. When mandates were implemented without earnings supple-
ES-2 14
ments, participants lost welfare benefits as they gained earnings, so these pro-grams did not usually raise family income or resources.
3. Time limits. One of the programs under study put time limits on families' eli-gibility for welfare benefits, restricting eligibility to a certain number ofmonths in a specified period. This program was a pilot welfare reform initia-tive implemented prior to 1996 under waivers of federal welfare rules. Until1996, cash welfare assistance was a federal entitlement that was available aslong as it was needed. The federal welfare law of 1996 sets a lifetime limit offive years on cash assistance receipt, but states may impose shorter limits orextend the time limits by using state funds. States may also exempt 20 percentof the caseload from the limits for hardship reasons. Once a family reaches thetime limit, federally funded cash benefits are terminated, but the family nor-mally remains eligible for food stamps, Medicaid, low-income child care as-sistance, and (where available) state-supported cash assistance. The programwith time limits combined them with mandatory employment services and asmall earnings supplement; the result was an increase in parental employmentbut only a modest increase in family income.
All the studies reviewed used a rigorous random assignment research design. Parentswere placed at random in either a program group, which had access to the new services and bene-fits and was subject to the new rules, or a control group, which received the benefits and wassubject to the rules that had previously existed in the locality of the study site or sites. In mostcases, members of the control group were eligible for cash assistance through Aid to Familieswith Dependent Children (AFDC), the cash welfare program in effect prior to 1996. Becauseparents were assigned to the groups at random, the average characteristics of families in the pro-
gram and control groups should not have differed systematically at the outset. The random as-signment method thus ensures that any differences between the two groups found during thestudy are due to the new program rather than to differences in the families' initial characteristicsor the general economic and social conditions that they experienced.
In surveys conducted two to four years after parents entered the studies examined here,children's school achievement, social behavior, and health were measured using parents' reportsand, in some studies, standardized tests or teachers' reports. To ensure the comparability of re-sults, we focused on a subset of measures that were similar across studies yet represented a widerange of outcomes for children that might be affected by welfare and work policies. Using thesemeasures, we conducted analyses for subsamples composed of single parents the great major-
ity of whom were women with children who ranged in age from approximately 3 to 9 whentheir parents entered the study. At the time at which school achievement, behavior, and healthwere measured, the children's approximate age range was 5 to 12. The findings for all the meas-ures of children's well-being and for the full samples can be found in the reports from the indi-vidual studies.'
2The Family Transition Program: Final Report on Florida's Initial Time-Limited Welfare Program (MDRC).2000. Dan Bloom, James Kemple, Pamela Morris, Susan Scrivener, Nandita Verma, Richard Hendra.
Impacts on Young Children and Their Families Two Years After Enrollment: Findings from the Child Outcomes
Study (U.S. Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evalua-tion and Administration for Children and Families; and U.S. Department of Education, Office of the Under Secretary
ES-3 15
The difference between the children in the program group families and those in the con-trol group families on a given outcome is referred to as the program's impact on that outcome.For each of the programs, we computed impacts and tested whether the impacts were statisticallysignificant (that is, unlikely to have occurred by chance). We also examined the patterns of im-pacts for the programs that shared each of the three features introduced above. "Increases" and"decreases" refer to the differences that emerged over time between the program and controlgroups. Our main findings follow.
The programs that included earnings supplements, all of which increasedboth parental employment and income, had positive effects on elementaryschool-aged children. All four programs that provided earnings supplementsled to higher school achievement. Some of the programs also reduced behav-ior problems, increased positive social behavior, and/or improved children'soverall health.
Adding mandatory employment services did not generally reduce thepositive effects of earnings supplements on children. The only program thatincluded mandatory employment services in addition to an earnings supple-ment increased parents' full-time employment but generally did not affectchildren's outcomes beyond having the same positive effects as the programdid when it was implemented with earnings supplements alone.
The programs with mandatory employment services, all of which boostedparental employment without increasing income, had few effects on chil-dren, and the effects were mixed in direction. These six programs had rela-tively few noteworthy effects on children. When impacts were found, the ef-fects were about equally likely to be positive as negative. The pattern of im-pacts appeared to be more closely associated with particular sites than withprogram characteristics like participation mandates.
The program with time limits, which led to an increase in parental em-ployment and a modest increase in income, produced few noteworthy im-pacts on children, and the impacts found did not suggest a consistent pat-tern of benefit or harm. Our knowledge base is smallest with regard to theimpacts of time limits because only one program had time limits, and this pro-gram combined them with mandatory employment services and a small earn-ings supplement. The program's few impacts on children were mixed: Healthimproved, but positive social behavior decreased.
These general conclusions are subject to the following caveats:
and Office of Vocational and Adult Education). National Evaluation of Welfare-to-Work Strategies. 2000. SharonMcGroder, Martha Zaslow, Kristin Moore, Suzanne LeMenestrel.
New Hope for People with Low Incomes: Two-Year Results of a Program to Reduce Poverty and Reform Wel-fare (MDRC). 1999. Johannes Bos, Aletha Huston, Robert Granger, Greg Duncan, Thomas Brock, Vonnie McLoyd.
Reforming Welfare and Rewarding Work: Final Report on the Minnesota Family Investment Program: Volume2: Effects on Children (MDRC). 2000. Lisa Gennetian, Cynthia Miller.
The Self-Sufficiency Project at 36 Months: Effects on Children of a Program That Increased Parental Employmentand Income (Social Research and Demonstration Corporation). 2000. Pamela Morris and Charles Michalopoulos.
ES-4 16
Although the effects of earnings supplements on children are encourag-ing, the improvements are modest when considered in the context of thesechildren's high levels of disadvantage. Even the programs with the mostbenefits to children left many families in poverty and many children at risk ofschool failure and behavior problems. These programs do not eliminate theneed for child-focused interventions and reforms that promote schoolachievement and reduce beh4vior problems.
The positive effects of earnings supplement programs on children weremost pronounced for the children of long-term welfare recipients. Forfamilies in which the parent had a long history of using welfare, the programswith earnings supplements improved children's development and increasedparental employment and family income.
The conclusions in this monograph are limited to preschool-aged andelementary school-aged children. Infants and toddlers, as well as adoles-cents, may be affected differently by the welfare reform approaches ex-amined here. Too few of the studies considered here specifically examinedchildren under age 3 for general conclusions to be drawn. For adolescents,however, two of the studies (one examining a program with an earnings sup-plement and another a program with a time limit) found decreases in schoolachievement and increases in behavior problems among adolescents.
Although the program features examined in this monograph are similarto those included in many programs that have been implemented bystates since 1996, they do not represent the full range of earnings supple-ments, participation mandates, and time limits currently in effect. Thepatterns from which these broad conclusions are drawn were observed in pro-grams in different geographic regions with different population characteristics,justifying some confidence that the findings will generalize across differentcontexts. Nonetheless, most of the studies were conducted prior to the passageof the 1996 federal welfare legislation, and their impacts could be different ina different macroeconomic or policy context. Moreover, while the policiesexamined here are representative of some of the state policies currently ineffect, policies that provide less generous supplements or impose morestringent mandates or time limits than those examined here may have differenteffects on children.
The welfare reforms initiated by the states and the legislated changes in the 1990s did notlead to one new welfare policy but to a variety of policies that continue to evolve. As welfarecaseloads decline, federal and state policies are generally being expanded to reach all working-poor families, regardless of their welfare status. The findings of this synthesis may guide policychoices that promote the development of children both in families receiving welfare and in otherlow-income families. Welfare reforms and antipoverty programs can have a positive impact onchildren's development if they increase employment and income, but increasing employmentalone does not appear sufficient to foster the healthy development of children. Children living inpoverty are at risk of low achievement, behavior problems, and health problems, so it is criticalthat policies affecting their families enhance children's well-being rather than leaving them at thesame level of deprivation and risk that they experienced under the former welfare system. Wehope that this analysis will help state and federal policymakers make informed choices that keepthe effects on children in focus as they design legislation that affects low-income parents.
ES-5 17
Chapter 1
Introduction
The Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA),
passed in 1996, was the culmination of several decades of efforts to promote work and reduce
long-term welfare receipt among single-parent families, the great majority of which are headed
by women. As a result of these efforts, the welfare system was not only transformed, but benefits
for working-poor families were expanded to reward work outside the welfare system through the
Earned Income Credit (EIC, the federal tax credit that supplements the earnings of low-income
families), publicly funded health insurance, and child care assistance. Whether promoting work
among low-income single parents helps or hurts children, and under what conditions, is the sub-
ject of this monograph. The monograph is a product of the Next Generation project, a collabora-
tion among researchers at the Manpower Demonstration Research Corporation (MDRC) and sev-
eral leading research institutions that is being funded by the David and Lucile Packard Founda-
tion, William T. Grant Foundation, and John D. and Catherine T. MacArthur Foundation.
Proponents of the policy changes have argued that parental employment benefits children
by increasing parents' self-esteem and providing children with positive role models. Skeptics
worry that greater work responsibilities may harm children by increasing parents' stress, reducing
the amount of time parents can spend with their children and monitoring them, and increasing the
amount of time children spend in low-quality child care arrangements. For parents unable to
maintain consistent employment, skeptics also worry that the loss of a financial safety net may
adversely affect children. In the context of increasing autonomy at the state level and the poten-
tial for welfare policy to evolve further at the federal level, understanding how children are af-
fected by different welfare policies is critical. This monograph seeks to advance our understand-
ing of how various welfare reform programs affect children by synthesizing evidence from
evaluations of 11 welfare and employment programs. A companion document' examines the ef-
fects of these and other programs on parental employment, welfare use, and income. The five
large-scale evaluations covered here are unusual in their experimental rigor and comprehensive
measurement of children's well-being.
The relevant policy background and an overview of this research synthesis are provided
in this chapter. In Chapters 2, 3, and 4, we analyze the effects on children's well-being of pro-
grams with earnings supplements, mandatory employment services, and time limits, respectively.
In those three chapters, we discuss how parenting, child care, and other family changes may ac-
count for the programs' effects on children. We also examine effects on the children of long-term
welfare recipients, on preschool-aged versus early school-aged children, and on boys versus girls.
Based on the results of the two studies that have examined adolescents, in Chapter 5 we explore
the possibility that the programs affect adolescents differently than children in other age groups.
Chapter 6 presents an overall summary of the research synthesis laid out in the preceding chap-
ters and discusses its implications for policy.
'Bloom and Michalopoulos, 2001.
18
I. Historical Background: Welfare Reform and Children
The 1996 federal welfare reform law introduced sweeping changes to the nation's systemfor supporting low-income families with children. During the prior six decades, Aid to Familieswith Dependent Children (AFDC) had guaranteed aid for economically deprived families withchildren. The new law eliminated AFDC, which was funded as an open-ended entitlement, andreplaced it with Temporary Assistance for Needy Families (TANF), which provided block grantsto states, introduced time limits on cash assistance, and imposed work requirements on recipi-ents. The law made other substantial changes affecting child care, the Food Stamp Program,Supplemental Security Income (SST) for children, and the Child Support Enforcement program,giving states numerous options for instance, the option to require work of parents with infants(children under 12 months old), to cap benefits so that payments do not increase if recipientshave additional children ("family caps"), and to require individuals to sign individual responsibil-ity plans. The most controversial of these changes was the institution of time limits on receipt offederal cash assistance. One of the studies examined in this monograph provides informationabout the effects of a time-limited welfare reform program on children.
These sweeping changes were the results of a more gradual process, begun in the 1960s,intended to push welfare recipients toward higher levels of employment-based self-support. In1967, Congress passed a law requiring parents who were receiving AFDC and who had no pre-school-aged children to register for work activities. Efforts to enforce work requirements variedwidely from one state to the next and were not taken seriously by most states until the early1980s. Changes in 1981 and in 1988 (following passage of the Family Support Act) sought toaccelerate states' efforts to promote employment and reduce welfare, but full implementationwas thwarted by the recession of the early 1990s. Efforts to understand the impact of these policychanges (such as mandatory employment services) on children's development gave rise to someof the evaluations included in this monograph.
In the four years prior to the 1996 legislation, the federal government granted waivers offederal AFDC rules to more than 40 states to allow them to experiment with program changes.The evaluations conducted under such waivers in Minnesota and Florida provide some of thedata used here. Waiver provisions varied widely among states. Some included earnings supple-ments designed to compensate recipients for lost welfare benefits as their earnings increased andthereby to encourage them to increase their work effort. Without supplements, welfare recipientstypically lost their benefits at about the same rate as they gained earnings, so their income did notincrease with their work effort. Many programs tested requirements that recipients participate inemployment-related activities or risk losing their benefits. Others tested time limits on welfarereceipt. Programs that included earnings supplements, mandatory employment services, and timelimits were tested in the studies examined here.
II. Program Features
The debate surrounding the passage of the 1996 welfare reform legislation has beenfraught with assumptions and predictions about the effects of the proposed reforms on children.Reform advocates foresaw many benefits to children of increased parental employment, whichthey believed would create positive role models, promote parents' self-esteem and sense of con-
-2- .19
trol, introduce productive daily routines into family life, and eventually foster career advance-ment and higher earnings on the part of both parents and children.2 In this view, children's devel-opmental needs are addressed indirectly but effectively by policies that promote employmentamong welfare recipients.
A very different view of the potential effects of welfare reforms on children stresses therole of family income and resources available to children. Armed with forecasts of dramatic in-creases in child poverty, critics of welfare reform focused on the likely detrimental effects onchildren's well-being of families' losing welfare benefits. Proponents were more optimistic thatas parents moved into jobs, their future earnings would elevate family income above the level ofwelfare benefits.
Welfare reform was also intended to encourage marriage and discourage out-of-wedlockchildbearing, both of which were expected to improve children's well-being. The preamble to the1996 PRWORA legislation identifies marriage as "an essential institution of a successful societywhich promotes the interests of children," posits that "responsible fatherhood and motherhoodare integral to successful child rearing and the well-being of children," and asserts that the "pre-vention of out-of-wedlock pregnancy and reduction in out-of-wedlock births are very importantGovernment interests." Limiting cash support for single parents and instituting caps on assistancefor women who have additional children while on welfare were aimed at discouraging out-of-wedlock childbearing. But these policies which were designed to affect marriage, paternal in-volvement, and family structure directly are not represented in enough studies to allow sys-tematic comparison of them.
In this monograph, we draw on studies of programs that were designed primarily to affectparental employment and income but that could have affected children indirectly. Specifically, weexamine the effects on children of programs that include three policy approaches currently used inmany state welfare programs: earnings supplements, mandatory employment services, and timelimits on welfare receipt. Each of these program features is designed to encourage work, reducewelfare use, and particularly in the case of earnings supplements increase income.
A. Earnings Supplements
Some welfare and antipoverty programs include strategies specifically designed to makework more financially rewarding than welfare. The labor market accessible to the working poor of-fers mainly low-skill, low-pay jobs that are often transitory, making welfare potentially more attrac-tive than work. Some welfare reform programs try to compensate for some of the shortcomings ofthe labor market by "making work pay" that is, by providing extra income and resources to re-cipients who are employed. Some such programs require full-time employment; others provideearnings supplements for any amount of work. Supplements are sometimes provided within thewelfare system by increasing the earnings disregard (the amount of earnings that is not counted asincome in calculating the amount of a family's welfare benefit) so that families can keep more oftheir welfare dollars when parents go to work. In other programs, earnings supplements are pro-vided outside the welfare system in the form of cash supplements and sometimes in addition
2Haskins, forthcoming, 2001.
-3-20
in-kind benefits such as child care or health care subsidies. Most states have already made enhancedearnings disregards a key component of their TANF policies. A few other states provide cash sup-plements through refundable tax credits. The studies examined in this monograph include policiesthat are comparable to the most generous disregards provided in many current state programs.
B. Mandatory Employment Services
Since the 1970s, welfare reform approaches have been designed to induce participation inwork-related activities or employment by making participation mandatory. The primary tool usedto enforce participation mandates is sanctioning, whereby a recipient's welfare grant is reduced ifshe or he does not comply with program requirements. These programs are designed to reducewelfare use and increase employment either by promoting parents' participation in job search andjob training or by requiring parents' participation in basic education, both with the long-term goalof increasing the employability of these often low-skilled workers. Today, virtually all states areusing such mandates in their attempt to reduce welfare use and increase parents' self-sufficiency.In some cases, the mandates are more stringent (with respect to the number of hours of work re-quired or the size of the sanction) than those in the studies examined here.
C. Time Limits
Until 1996, cash welfare assistance was a federal entitlement that was available to fami-lies as long as they met the eligibility requirements. The federal welfare law of 1996 sets a life-time limit' of five years on cash assistance receipt, but states may shorten or extend the limits byusing state funds. States may also exempt 20 percent of the caseload from time limits for hard-ship reasons. Once a family reaches the time limit, federally funded cash benefits are terminated,but the family normally remains eligible for Food Stamps, Medicaid, low-income child care as-sistance, and (where available) state-supported cash assistance. Time limits are intended to re-duce welfare dependence, encouraging parents to work in order to support their families. Morethan 40 states have established limits on the receipt of cash assistance, ranging from 21 to 60months.
HI. The Relation Between Program Features and Children's Outcomes
To further our understanding of how policy features such as earnings supplements, man-datory employment services, and time limits affect children, we draw on extensive research onthe link between parental employment, welfare dependence, and income on the one hand andchildren's outcomes on the other. The resulting conceptual model is presented in Figure 1.1. Acentral idea behind this model is that policy changes can affect children indirectly throughchanges in resources (for instance, child care, housing, learning materials, and food) and throughchanges in children's socialization experiences (their family functioning and relationships withparents). Examining such intermediate outcomes, or possible mediators of programs' effects onchildren, helps us understand not only whether a policy affects children but also how it does so.
3Lifetime limits restrict the number of months in the recipient's lifetime that she or he can receive welfare bene-fits. Fixed-period time limits, in contrast, restrict the number of months of benefits over a shorter, specified periodfor example, to 24 months in any 60-month period. The time-limited program examined in this monograph includes afixed-period limit rather than a lifetime limit.
21-4-
How Welfare and Work Policies Affect Children
Figure 1.1Mechanisms Through Which Welfare and Employment Policies Might Affect Children
Program Parents' Economic Intermediate Children'sFeatures Outcomes Outcomes Outcomes
Earnings supplements Employment Resources AchievementBooks and toys
Mandatory employmentservices
Family income Child care Positive behavior
Welfare receipt Socialization Behavior problems
Time limits Parents' well-beingParenting behavior Health
22-5-
Below we review the literature on the relations between the primary targets of these interventions(parental employment, income, and welfare use) and children's development, paying particularattention to the possible dependency of these effects on the age and gender of the child.
A. Effects of Maternal Employment
The effects of maternal' employment on children depend on the characteristics of themother's job, the extent to which family resources increase, the mother's psychological well-being, and the quality of the child care, youth programs, and neighborhood to which the child isexposed. For low-income families headed by single mothers, in particular, the associations be-tween maternal employment and children's cognitive and social development tend to be posi-tive.' But it appears that much, if not all, of this difference stems from differences between em-ployed and unemployed mothers in their demographic attributes, skills, personalities, and child-rearing practices rather than from their employment status per se.' Moreover, holding highlyroutinized jobs that pay very low wages and afford little autonomy appears to have negative ef-fects on mothers' emotional well-being and, in turn, on children's development.' Unpredictableand unconventional work hours, which are characteristic of many low-wage jobs, may make itdifficult for parents to combine work and family responsibilities.,Maternal employment may havemore positive effects on children when mothers believe their children will not suffer as a resultthan when it conflicts with their beliefs about what is best for their families.'
B. Effects of Family Income
In studies conducted in the United States, poverty has been found to have small but consis-tently negative effects on children's development.' Unsurprisingly, persistent and deep poverty hasbeen shown to be more detrimental to children than transient poverty.'" Family income may influencechildren through both the resource and socialization pathways in Figure 1.1 affecting the resourcesparents can provide to their children and influencing parental stress and parenting behavior." In thefew longitudinal studies that have been conducted, family income consistently predicts children's aca-demic and cognitive performance, even when other family characteristics are taken into account.'2Children from low-income families also have more behavior and health problems than those frommore affluent families do.' Family characteristics associated with poverty account for thesedifferences in some studies."
4Because the vast majority of the single parents in the studies included in this monograph were mothers, here wereview research that focuses specifically on the effects of maternal employment on children.
5Harvey, 1999; Vandell and Ramanan, 1992; Zaslow and Emig, 1997.6Zaslow, McGroder, Cave, and Mariner, 1999.7Moore and Driscoll, 1997; Parcel and Meneghan, 1994, 1997.8Jackson, 1993; NICHD Early Child Care Research Network, 1998a.9Duncan, Brooks-Gunn, and Klebanov, 1994; Duncan and Brooks-Gunn, 1997; Mayer, 1997; McLoyd, 1998.10Duncan et al., 1994; Bolger, Patterson, Thompson, and Kupersmidt, 1995."Bradley and Caldwell, 1984; Smith, Brooks-Gunn, and Klebanov, 1997; Sugland, Zaslow, Smith, Brooks-
Gunn, Moore, Blumenthal, Griffin, and Bradley, 1995; McLoyd, Jayartne, Ceballo, and Borquez, 1994.12Duncan and Brooks-Gunn, 1997."Duncan and Brooks-Gunn, 1997; Klerman, 1991; Korenman and Miller, 1997.14Duncan and Brooks-Gunn, 1997.
23-6-
C. Effects of Welfare Receipt
Some have argued that income from welfare is less beneficial to children than otherforms of family income because it carries a stigma. The research evidence does not consis-tently support or refute this hypothesis. Many studies found no relation between welfare re-ceipt and children's cognitive and social development once demographic and family character-istics are taken into account; in rare cases, positive relations were found." Other studies re-vealed that children in families receiving welfare have lower-quality home environments,'lower academic achievement," and lower completed schooling' than children in other poorfamilies. It is possible, however, that people receiving welfare have fewer material resourcesand assets than do other low-income families, which might explain these differences. More-over, entry into and out of welfare programs is often associated with other transitions andchanges (such as job loss or entry, parents' separating or acquiring new partners, and changesin child care arrangements) that affect parents' and children's well-being. Several studies ob-served higher levels of behavior problems (as reported by mothers) among children whosefamilies had recently made a transition into welfare" and among children whose families hadrecently left welfare" than among children whose families had not recently changed status.
D. Differences by Age
Maternal employment and family income may have more profound effects on youngchildren than on children who have reached school age. Developmental theories suggest thatinfants and preschool-aged children are more sensitive than older children to separation fromtheir parents. In addition, one study indicates that poverty during the preschool years predictscognitive development and educational attainment better than does poverty during middlechildhood or adolescence.'
The available data suggest that the effects of maternal employment on young children'scognitive and language skills depend on the quality of the child care provided while the motheris working, which may in turn be influenced by family income. The cognitive and languageskill development of children in low-income families benefits from high-quality care as com-pared with low-quality care.' In addition, formal, center-based child care is more beneficial tocognitive development than home-based care when the two are of comparable quality.'
15Butler, 1990; Haveman and Wolfe, 1995; Levine and Zimmerman, 2000; Ratcliffe, 1996; Yoshikawa, 1999;Zill, Moore, Smith, Stief, and Coiro, 1995.
16Moore, Morrison, Zaslow, and Glei, 1994; Smith and Brooks-Gunn, 1994.'Smith and Brooks-Gunn, 1994; Hofferth, Smith, Mc Loyd, and Finkelstein, 2000.18Duncan and Yeung, 1995.19Smith and Brooks-Gunn, 1994; Moore et al., 1994.26Hofferth et al., 2000. .
21Duncan and Brooks-Gunn, 1997.22Burchinal, Roberts, Riggins, Zeisel, Neebe, and Bryant, 2000; NICHD Early Child Care Research Network,
2000; Peisner-Feinberg, Burchinal, Clifford, Cu lkin, Howes, Kagan, Yazsjian, By ler, Rustici, and Zelazo, 1999;Ramey, Campbell, Burchinal, Skinner, Gardner, and Ramey, 2000; Shonkoff and Phillips, 2000.
23 NICHD Early Child Care Research Network, 2000.
24-7-
The effects of nonmaternal child care on health and social behavior are more mixed thanthose on cognitive development. Infants who are placed in group child care arrangements havehigher rates of contagious illnesses than do infants who are cared for at home, although the dif-ference declines by age 3.24 The effects of child care (whether of high or low quality) on chil-dren's social behavior, however, are much less consistently positive or negative.'
Once children begin school, the quality of child care may play a less critical role in theacquisition of academic skills because children are receiving instruction in school. At this stage,the quality of out-of-school activities and community supports becomes an important influenceon the development of positive behavior and behavior problems.' Adolescents may be morecognizant than younger children of the characteristics and value of their parents' work away fromhome; hence, they may benefit more from the positive role modeling of maternal employment.Moreover, older children may simply need less time with their parents than do their youngerpeers. On the other hand, employment may make it more difficult to monitor older children's ac-tivities, especially as they become independent of adult supervision in adolescence. There issome evidence that for adolescent children maternal employment, in conjunction with low levelsof monitoring and communication, is related to delinquency,27 low educational attainment,' andlow well-being."
E. Differences by Gender
Theory and research in developmental psychology suggest that maternal employment af-fects girls and boys differently. The extensive literature on how parental employment affectschildren is guided by the theory that children use their same-sex parent as a model of their ownfuture employment possibilities." Because most participants in welfare programs are mothers,changes in maternal employment could have more positive modeling effects for daughters thanfor sons. On the other hand, one consequence of increased maternal employment may be thatgirls are asked to perform more housekeeping and child care tasks, especially in low-incomefamilies that cannot afford paid help. A moderate number of such tasks may promote girls' de-velopment, but extensive adult responsibilities may interfere with their school achievement andlead to other problems.
A wide range of evidence indicates that boys are more vulnerable to problems in school,behavior problems, and poor health than are girls.' Hence, boys may show more negative effectsof parental employment than girls, particularly if their parents are in programs that increase stressor reduce monitoring. But because many parents are aware of boys' greater likelihood of beingaggressive and disobedient, they may exert more energy and expend more resources on prevent-
24NICHD Early Child Care Research Network, forthcoming, 2001.25NICHD Early Child Care Research Network, 1998b; Peisner-Feinberg et al., 1999.26Marshall, Garcia, Marx, McCartney, Keefe, and Ruh, 1997; Pettit, Laird, Bates, and Dodge, 1997; Pettit,
Bates, Dodge, and Meece, 1999.27Sampson and Laub, 1994.28Duncan and Yeung, 1995.29Crouter, Bumpass, and McHale, 1999.30Huston, 1983.31Golombok and Fivush, 1994.
25-8-
ing behavior problems in their sons than in their daughters. For example, ethnographic observa-tions of families in the New Hope program in Milwaukee, Wisconsin, suggest that parents mayhave invested their increased income in funding their sons' (as opposed to daughters') after-school activities and meeting their sons' needs because they worried about boys' behavior prob-lems increasing."
IV. Examining Children in Random Assignment Studies
In each of the studies examined in this monograph, families were assigned at random(through a lottery-like process) to either a program group or a control group. The program groupwas subject to the rules and benefits of the new program, while the control group was subject tothe prior program (usually the AFDC program in operation during or prior to that period). Be-cause the two groups did not differ systematically at the beginning of the study, any differencesbetween them found during the study can be reliably attributed to differences between thegroups' experiences in their respective programs.
Examining programs' effects on children in the context of a random assignment study of-fers several advantages over the research just reviewed on the effects of employment, welfarereceipt, and income on children. First, poor families probably differ from nonpoor families andworking families from nonworking families on dimensions other than income and employmentstatus. For example, low-income mothers may be more depressed on average than mothers withhigher incomes. Therefore, the fact that children in higher-income families perform better inschool could be explained either by their parents' higher incomes or by their parents' lower levelsof depression. Unfortunately, in the research reviewed above, it is very difficult to disentangle thepositive effects of employment and income from such unmeasured differences between families.In the studies examined in this monograph, in contrast, the random assignment design ensuresthat any systematic differences in children's outcomes can be confidently attributed to the pro-gram and not to unmeasured characteristics. The studies described above also do not examine theeffects on children of changes in parental employment, welfare receipt, and income over time butinstead compare children in families with different a priori levels of employment, welfare receipt,and income. As a result, even if the differences between poor and nonpoor children were due topoverty, on the basis of these studies we cannot know if increasing family income improves thelives of children in poor families to the same degree as in nonpoor families. Because the pro-grams examined here were targeted at employment, welfare receipt, and income, any differencesin children's outcomes between the program and the control groups are likely to be related to thechanges in these economic outcomes for parents.
V. The Studies
This monograph presents results for children whose parents participated in the five stud-ies described in Box 1.1, in which a total of 11 welfare and employment programs were evalu-ated. Each program included a variety of features. We classify the programs on the basis of thethree features already discussed: earnings supplements, mandatory employment services, and
32Bos, Huston, Granger, Duncan, Brock, and McLoyd, 1999.
26-9-
time limits. By looking across programs that differed with respect to these features, we can makeinferences about the effects of particular policy approaches on children.
How Welfare and Work Policies Affect Children
Box 1.1
Studies Examined in This Monograph
The Next Generation project analyzes data from five program evaluations, building on their re-search designs, outcome measures, and impact analyses. The evaluations and the organizationsthat conducted them are listed below.
Florida's Family Transition Program was evaluated by MDRC under contract to the Florida De-partment of Children and Families.
The Minnesota Family Investment Program was evaluated by MDRC under contract to the Min-nesota Department of Human Services.
The National Evaluation of Welfare-to-Work Strategies is being conducted by MDRC undercontract to the U.S. Department of Health and Human Services. The Child Outcomes Study,which examines program impacts on young children, is being conducted by Child Trends undersubcontract to MDRC.
The New Hope program is being evaluated by MDRC under contract to the New Hope Project,Inc., in collaboration with researchers from Northwestern University, the University of Texas atAustin, the University of Michigan, and the University of California at Los Angeles.
The Self-Sufficiency Project was conceived by Human Resources Development Canada. Theproject is being managed by the Social Research and Demonstration Corporation (SRDC) andevaluated by SRDC and MDRC.
A. The Minnesota Family Investment Program (MFIP)33
MFIP was begun as a pilot program in 1994 and implemented in seven urban and ruralcounties in Minnesota until 1998." The child study included single-parent families who were ap-plying for or currently receiving welfare. MFIP combined mandatory employment services and"make-work-pay" supplements. These earnings supplements were provided for either full- orpart-time work. Children were assessed three years after parents' enrollment in the program. Theevaluation tested two pilot programs," and welfare recipients were randomly assigned to one ofthree groups the control group or one of the two programs:
33 Gennetian and Miller, 2000; Knox, Miller, and Gennetian, 2000; Miller, Knox, Gennetian, Hunter, Dodoo, andRedcross, 2000.
341n 1998, a modified version of MFIP was implemented statewide. Statewide MFIP includes a less generousearnings supplement and a more stringent participation mandate than the pilot program and a time limit on cash as-sistance receipt.
35Strictly speaking, MFIP was a single program encompassing the full program and a variant thereof, but for thepurposes of this monograph the full program and its variant are referred to as separate programs.
-10- 2 7
Full MFIP combined all the features of the MFIP program into a single pack-age. The most important features of this package included (1) an earnings sup-plement that allowed working welfare recipients to keep more of their incomewhen they went to work (also known as an increased earnings disregard); (2)mandatory employment services requiring long-term welfare recipients to par-ticipate in employment or training activities unless they were working morethan 30 hours per week or had a child under 12 months old; (3) child carepayments made directly to the provider; and (4) a streamlining of the rules fordisbursement of cash assistance in which (a) the AFDC, Food Stamp, andFamily General Assistance programs were consolidated and Food Stampswere "cashed out" (that is, the Food Stamps' dollar value was included in thewelfare check) and (b) the eligibility rules for single-parent and two-parentfamilies were equalized.
MFIP Incentives Only included all the features of the Full MFIP programexcept mandatory employment services.
B. The Self-Sufficiency Project (SSP)36
This program was launched in 1992 and operated in two provinces in Canada (NewBrunswick and British Columbia) until 2000. SSP took a pure make-work-pay approach, offeringa generous earnings supplement for full-time work (at least 30 hours per week) for up to threeyears. The earnings supplement was a monthly cash payment available to single-parent welfarerecipients who had been in Canada's welfare program for at least one year and who chose toleave welfare for full-time work within a year of being offered the supplement. The amount ofthe supplement was calculated as half the difference between a recipient's earnings and an earn-ings benchmark set such that a parent with a full-time minimum-wage job would roughly doubleher income if she received the supplement. Data on children were collected three years after par-ents were enrolled in the program.
C. The New Hope Program'
New Hope operated in two low-income areas of Milwaukee, Wisconsin, from 1994 until1998. Because the program was targeted at low-income families who were willing to work fulltime, the study included both welfare recipients and other low-income parents. New Hope in-cluded various make-work-pay strategies. As in SSP, parents who worked full time (at least 30hours per week) were eligible for a cash supplement that was intended to bring their income tothe poverty line. They could also elect to receive child care and health insurance subsidies. In thisprogram, then, in-kind benefits were available to families in addition to the cash supplement. Tohelp families take advantage of the program's benefits, New Hope also provided intensive casemanagement and, for parents who did not find full-time work, short-term community servicejobs. Assessments of children were made two years after parents were randomly assigned eitherto the program or the control group.
36Michalopoulos, Card, Gennetian, Harknett, and Robins, 2000; Morris and Michalopoulos, 2000.37Bos et al., 2000.
28
D. The National Evaluation of Welfare-to-Work Strategies (NEWWS)38
The part of NEWWS examined here' included six programs in three sites (Atlanta, Geor-gia; Grand Rapids, Michigan; and Riverside, California) that operated in the early to mid 1990sunder the federal Job Opportunities and Basic Skills Training (JOBS) Program, which precededTANF. Like TANF, the primary objective of these programs was to reduce single parents'welfare use and increase their employment. The programs required welfare recipients to partici-pate in basic education or employment-related activities as a condition of receiving welfare.Families who failed to meet the participation requirements could receive sanctions.
For purposes of the evaluation, each of the three sites examined here operated both aprogram stressing job search as a first activity and a program stressing basic education as afirst activity; all three sites served welfare recipients. This design permitted a direct, side-by-side comparison of the two approaches. Information on children was collected as part of theChild Outcomes Study' two years after parents were randomly assigned to one of the two pro-grams at their site.
Job-search-first programs. These programs required most participants tolook for work immediately, usually by attending a "job club" that lasted one tothree weeks. Most people who completed job search without finding a jobwere then enrolled in short-term adult basic education, vocational training, orwork experience.
Education-first programs. These programs initially placed participants ineducation and training programs (usually adult basic education or vocationaltraining) to increase welfare recipients' "human capital" (knowledge andskills) before they attempted to move into employment.
E. Florida's Family Transition Program (FTP)41
FTP operated in Escambia County (which includes the city of Pensacola) between 1994and 1999.42 The program combined a small earnings supplement with a time limit on the receiptof welfare benefits. The supplement, which was provided through the welfare system as an en-hanced earnings disregard, provided only a small amount of additional income to families whotook advantage of it because the welfare benefit levels in Florida were quite low. Receipt of cashassistance was limited to 24 or 36 months (depending on the parents' level of disadvantage) inany 60-month period. However, exemptions were granted to parents who became disabled or in-capacitated while receiving assistance (these recipients' welfare clocks" were stopped so thatnot all the months of welfare benefits they received were counted). Parents in FTP were also pro-
38Freedman, Friedlander, Hamilton, Rock, Mitchell, Nudelman, Schweder, and Storto, 2000; Hamilton, 2000;McGroder, Zaslow, Moore, and LeMenestrel, 2000.
39The programs discussed here are among the 11 programs operating in seven sites that were evaluated as part ofNEWWS (Freedman et al., 2000).
40McGroder et al., 2000.41Bloom, Kemple, Morris, Scrivener, Verma, and Hendra, 2000.42FTP was a precursor of Florida's WAGES (Work and Gain Economic Self-Sufficiency), a statewide program
that operated from 1996 to 2000 and shared many of FTP's features but differed from it in key ways.
-12- 29
vided with an array of services to help them find work in which they were required to participateor risk facing sanctions. Several child-focused mandates one requiring parents to ensure thattheir children were immunized and the other that their children were attending school regularly(with stipulations about regular parent-teacher contact) were also imposed. Unlike the otherstudies examined in this monograph, the control group in FTP was also subject to some manda-tory employment services; however, the mandatory services in the FTP group were more com-prehensive and monitored participants more closely. Data on children were collected four yearsafter parents' enrollment in the FTP or control group, making FTP's follow-up period the longestof all those in the studies examined here.
VI. Comparisons Across Programs
As the thumbnail program descriptions show, few of the welfare reform approaches testedin the studies included only one of the program features described earlier. Rather, because poli-cymakers often design programs with multiple goals in mind, most of the programs "packaged"two or more of the features together. Those that included only one of the three features examinedhere provide the most compelling information about that particular component's effect on chil-dren. Those with multiple features cannot provide direct evidence about the effects of single fea-tures because features may interact in affecting children's outcome. For example, earnings sup-plements may have different effects on children when combined with mandatory employmentservices than when offered alone. However, programs with multiple features can shed light onhow different features work together in influencing children's outcomes and how the addition ofa specific feature to the mix influences the effects of particular policy approaches. Finally, pro-grams that have been studied using a multiple-research-group design (such as MFIP) allow for arigorous assessment of the effects of adding a specific feature to a set of other features (in thecase of MFIP, adding mandatory employment services to an earnings supplement).
Table 1.1 characterizes the 11 programs under study with respect to whether they in-cluded earnings supplements, mandatory employment services, and time limits to illustrate whatcan be learned from comparing the effects of the different programs. The programs in NEWWSprovide the best information on the effects of mandatory employment services alone because theyprovide mandatory employment services without earnings supplements or time limits. The otherfive programs all included earnings supplements. Of the five, only three programs offered a gen-erous earnings supplement in the absence of mandatory employment services and time limits,although they differ in other respects (for instance, in the extent of case management and in theuse of in-kind benefits in supplement packages). The Full MFIP program affords information onthe effects of a generous earnings supplement combined with mandatory employment servicesand, when compared with the MFIP Incentives Only program, provides information on the ef-fects of adding a mandate to a generous earnings supplement. Finally, only one program withtime limits FTP has been studied to date, making our conclusions about the effects of timelimits necessarily tentative. Moreover, because in FTP time limits were combined with manda-tory employment services and a small earnings supplement, we can analyze the effects of timelimits only when they are combined with other features.
30-13-
How Welfare and Work Policies Affect Children
Table 1.1
Features of Each Program, by Study
MandatoryEarnings Employment
Study Supplements Services Time Limits
MFIPFull MFIPMFIP Incentives Only
SSPNew HopeNEWWS
Atlanta Job-Search-FirstAtlanta Education-FirstGrand Rapids Job-Search-FirstGrand Rapids Education-FirstRiverside Job-Search-FirstRiverside Education-First
FTP
Although there has been considerable discussion about how the 1996 federal legislationaltered welfare policy, many of the features that have been incorporated into programs since 1996were already being tested in these studies before 1996. Thus, these studies are a powerful tool forunderstanding how some post-TANF programs might be affecting children.
VII. Time Periods of the Studies
All the studies examined programs that began to be implemented prior to the 1996 wel-fare law. Nevertheless, there is some important variation in the study periods and therefore in theeconomic and policy conditions under which the studies took place. The earliest programs arethose analyzed as part of the NEWWS evaluation. In these programs, families were randomlyassigned to program and control groups between 1991 and 1993, and children were assessed twoyears later (between 1993 and 1995). Thus, the entire study period predated PRWORA's passagein 1996. The periods of the MFIP, New Hope, and FTP studies, in contrast, overlapped withPRWORA's passage, with families' being randomly assigned between 1994 and 1995 and chil-dren's being followed up in or after 1996.
The evaluation of SSP, which operated in Canada, spanned both of these periods, withfamilies' being randomly assigned between 1992 and 1995 and children's being assessed be-tween 1995 and 1998. As in the United States, the Canadian government passed legislation in1996 that turned welfare into a federally funded block grant, giving localities considerable discre-tion. However, unemployment was considerably higher in the provinces in which SSP was as-sessed than it was in the United States, although the local economic conditions improved slightlyover the study period.
-14- 3 1
This discussion underscores the range of economic and policy conditions under whichthese studies were conducted. Consistent findings across these studies would support strongerconclusions about the effects of particular program features and generalization of the findings toa wider range economic and policy contexts than would otherwise be warranted. Inconsistentfindings across studies might be attributable either to differences in the packages of features theprograms offered or to differences in the conditions under which the studies occurred, therebycomplicating the conclusions that can be drawn from the cross-study comparison. Nevertheless,the relatively short period of time over which these studies took place reduces the possibility thatdrastic differences in economic conditions underlie differences in the studies' findings.
VIII. Sample, Measures, and Analysis Strategy
A. Sample
We focus on children of single parents (primarily single mothers) who were of preschoolage or early school age at the time of their parents' random assignment (approximately 3-9 yearsat random assignment).' These children had reached middle childhood (approximately 5-12years) by the end of the follow-up period, from two to four years later. This age group was cho-sen as the basis for the cross-study analysis to maximize the comparability of the samples acrossthe 11 programs and because this age group was the subject of a set of detailed questions aboutsocial behavior and well-being in all the studies. Thus, the data collected allow for a fair com-parison of program impacts across studies.
Focusing on children of preschool and early school age has additional advantages. First,the most reliable assessments of children's functioning available in developmental psychologyare designed for children in this age range. Second, there is reason to believe that this age groupcontains the children most likely to be affected by changes in welfare and employmentprogramsbecause they are young enough to react negatively to maternal separations and to be placed innonmaternal care. At the same time, research has suggested that the youngest of these children(those aged 3-5) may benefit the most from increases in income.
In all the studies, some data were collected on other age groups of children as well, but inmost cases the information was too limited to allow for systematic analysis. In Chapter 5, webriefly discuss the effects on adolescent children found in the two studies for which more de-tailed information was collected and analyzed. Future research will analyze the differential ef-fects of welfare and employment programs on younger and older children in greater detail.
43Although we focus on preschool-aged and early school-aged children in many of the analyses in this mono-graph, the age groups are not identical across the 11 programs discussed. The footnotes in the figures specify thesubgroups of children for whom data are presented. In cases in which the age group included children beyond thosewho were aged 3-9 at random assignment, analyses were conducted to determine if the effects are comparable whenthe sample of children was limited to those in the narrower age range. When the effects are indeed comparable, theresults for the larger age group are shown in the figure to include the largest possible sample of children.
2-15-
B. Measures
To increase the comparability of the results found in the different studies, we examined asubset of measures that was similar across studies but that represented the range of children'soutcomes that might be affected by welfare and employment policies. The findings on the full setof measures of children's well-being can be found in the reports on the individual studies." Herewe examine effects in three broad categories: cognitive or achievement outcomes, social behav-ior, and health.
Children's cognitive outcomes include school achievement as rated by parents and teachersas well as children's test scores. Test scores can provide information about children's actual knowl-edge of a particular area, while parents' and teachers' reports provide information on how childrenare performing in school, which likely reflects both their cognitive abilities and their engagement inschool. Some of the studies included only reports by parents, while others included informationfrom more than one source.
Both positive and negative aspects of children's social behavior are also examined. Interms of behavior problems, the analyses presented here focus on externalizing behavior (chil-dren's "acting out" and engagement in overtly negative interactions such as yelling at and fight-ing with adults and peers) rather than internalizing behavior (depression and anxiety). We devoteour attention to externalizing rather than internalizing behavior both because the former is moreeasily and accurately assessed by parents and teachers and because it has been shown to be influ-enced more than internalizing behavior by child-focused interventions.' We also look at chil-dren's positive social behavior as measured by the extent to which children are helpful and coop-erative in their interactions with others.
Finally, we analyze parents' ratings of children's general health. In most studies, theseratings are responses to a single question.
In general, test scores and reports by adults other than parents are more reliable methodsof assessing the effects on children of programs aimed at parents than are parents' reports. Be-cause program group parents' perceptions of their children may be influenced by their experi-ences in the program, differences between their reports and those of control group members mayreflect those differences in perception rather than or in addition to differences in children's actualfunctioning. For example, parents who are stressed by being employed may perceive their chil-dren as more poorly behaved than parents who are less stressed, even if the behavior of the chil-dren in the two groups is the same. Similarly, parents who are working may be more aware oftheir children's health problems than parents who are not because employed parents might haveto miss work when their children are sick. These caveats are not intended to imply that parents'reports are not useful sources of data. In fact, to the extent that they translate into parents' behav-ior toward children, they may be very accurate at predicting long-term outcomes for children.However, in assessments of the effects of these programs on children, teachers' reports and chil-dren's test scores should be weighed more heavily than parents' reports.
44Bloom, Kemple, et al., 2000; Bos et al., 1999; Gennetian and Miller, 2000; McGroder et al., 2000; Morris andMichalopoulos, 2000.
45Yoshikawa, 1995.
-16- 33
C. Analysis Strategy
Each of the studies examined in this monograph used a random assignment research de-sign to measure the effects, or impacts, of the program(s) on various measures of outcomes forparents and children. We computed the impacts on the measures described in the previous sec-tion. In reporting impacts, we often refer to the differences that emerged over time between thegroups as "increases" or "decreases."
For each impact, we conducted a statistical test' to determine if the impact was statisti-cally significant. Some differences between the program and control groups might arise fromchance. A test of statistical significance indicates when an impact is unlikely to be due to suchchance differences between groups. Unless otherwise noted, every "impact" and "effect" men-tioned in this document was determined to be statistically significant.
The programs' impacts are represented here in terms of effect sizes. A larger effect size(whether positive or negative) corresponds to a larger effect. Effect sizes are computed by dividingthe difference between the average program and control group outcomes on a particular measure bythe standard deviation of the control group members' outcomes on that measure.' (The standarddeviation of a set of observations captures their degree of "spread," or how variable they are relativeto the average outcome.) The effect size is a yardstick for measuring impacts that can be used tocompare effects even when they are measured on different scales or in research samples that havedifferent standard deviations. For example, effect size analysis allows us to determine whether a 5-point difference between the program and control groups in the percentage of children performingbelow average in school is similar in magnitude to a 10-point difference between the program andcontrol groups' average scores on a test graded on a scale from 1 percent to 100 percent.
Random assignment designs make it possible to attribute systematic differences betweenthe program and control groups to the different programs the two groups experienced. In thismonograph, however, we compare the effects of the different program approaches across studies.Inferences based on cross-study comparisons are more tenuous than those made within a particu-lar study because factors other than the differences in program models may explain why the find-ings from two studies differ. Wherever possible, we attempt to address this problem by selectingsimilar subgroups for comparison. However, this technique is far from perfect, and other factorsmay still confound comparisons across studies.
46Specifically, a two-tailed test was performed for each difference in outcomes between the program and controlgroups to determine whether it was statistically significant. In the studies in which more than one child per familywas included in the study sample, the statistical tests were adjusted to account for the relation between siblings. Dif-ferences between subgroup impacts were also tested for statistical significance.
47The control group standard deviation is used because the program may affect the degree of spread in the pro-gram group (for example, the program may help some children and hurt others, thereby increasing the standard de-viation). Using the control group standard deviation allows one to examine the difference as a function of the varia-tion in the control group's environment.
3 4-17-
Chapter 2
Effects on Children of Programs with Earnings Supplements
Earnings supplements are designed to encourage work and reduce poverty. Given thatchildren are negatively affected by poverty, increasing income through earnings supplementsmight be expected to benefit children. But to receive an earnings supplement parents have towork, and some of the programs examined here required that parents work full time. It is there-fore possible that the increases in employment produced by earnings supplements lead to nega-tive effects on children that undermine the positive effects of higher family income. This chapter,which is based on a reanalysis of the data from three of the five studies introduced in Chapter 1(MFIP, SSP, and New Hope'), speaks to these issues by analyzing the effects on children of earn-ings supplement programs.
I. Effects on Parents' Economic Outcomes
Because programs that provide earnings supplements are targeted at adult behavior, they aremost likely to affect children through changes in economic outcomes for parents. Here we brieflydescribe the effects of the programs with earnings supplements analyzed in this monograph withrespect to three economic outcomes: employment, welfare receipt (the proportion of families re-ceiving welfare), and income. This summary is based on a companion document2 that examines theeffects of these and other programs on parents' economic outcomes in greater detail.3
Earnings supplement programs are intended to increase employment,and those examined here appear to have achieved this goal.
The programs with earnings supplements generally increased employment, although pri-marily for long-term welfare recipients. Whether the increase is in full-time employment (30 ormore hours per week) or part-time employment depends on the nature of the supplement. For ex-ample, because the MFIP Incentives Only program rewarded part-time work more than full-timework (families in which parents worked part time actually experienced a larger income boostfrom the program than families in which parents worked full time), the program increased pri-marily part-time employment for long-term welfare recipients. SSP, in contrast, increased full-time more than part-time work, because its supplements were contingent on full-time employ-ment. The Full MFIP program, which included both an earnings supplement and a mandate forfull-time work, also increased full-time employment for long-term welfare recipients.
The effects of the earnings supplement programs examined here on wel-fare use depended largely on the way in which the supplement was pro-vided.
'For the MFIP evaluation, see Gennetian and Miller, 2000, and Knox, Miller, and Gennetian, 2000; for the SSP evalua-tion, see Morris and Michalopoulos, 2000; and for the New Hope evaluation, see Bos et al., 1999, and Huston, Duncan,Granger, Bos, McLoyd, Mistry, Crosby, Gibson, Magnusson, Romich, and Ventura, forthcoming, 2001.
2Bloom and Michalopoulos, 2001.3Note that in describing the effects of SSP, Full MFIP, and MFIP Incentives Only, we have focused on long-
term welfare recipients. Later in this chapter, we discuss outcomes for families in these studies with shorter welfarehistories.
-18- 35
The two programs that provided earnings supplements inside the welfare system by in-creasing the earnings disregard (Full MFIP and MFIP Incentives Only) increased both welfarereceipt and welfare payments as they increased employment for long-term welfare recipients.This is because more working families qualified for welfare in these programs than would havequalified under the AFDC system. One of the two programs that provided supplements outsidethe welfare system in the form of cash and/or in-kind benefits (SSP), in contrast, reduced welfarereceipt, but both programs that provided supplements outside the welfare system (SSP and NewHope) unsurprisingly increased receipt of assistance in the ;Conn of supplements.
Most importantly, the earnings supplement programs increased incomeand reduced poverty.
Because these programs supplemented the earnings of families who went to work, the incomelevels of families in the program groups in the studies examined here were typically much higher thanthose of families in the control groups. For example, long-term welfare recipients in Full MFIP hadincome levels about $1,200 per year higher on average than those in the control group.
Because the programs with earnings supplements increased cash trans-fers to families, these programs cost the government money.
Earnings supplements can increase earnings and income but at a cost to the government.For example, the net cost of MFIP per family for services, cash assistance, and Medicaid wasabout $2,000 per year for single-parent long-term recipients, and SSP's net cost per family wasabout $450 per year all of it spent on cash assistance because the program did not offer specialservices. New Hope was the most expensive of the three programs partly because it provided amore comprehensive package of services and partly because it generated smaller welfare savings(some families in the study were not welfare recipients to begin with). The net cost of New Hopeper family was about $4,000 per year.
II. Effects on Children
We first consider the effects on children of the three earnings supplement programs thatdid not include any form of mandatory employment services (MFIP Incentives Only, SSP, andNew Hope). All three programs gave parents a supplement if they went to work by providingcash, and New Hope offered additional family supports (child care and health insurance subsi-dies) to parents who worked.
All three programs shared one feature: the provision of a generous earnings supplement.However, in other ways these three programs were quite different for example, with respect towhether the earnings supplement was tied to full-time work. In the MFIP Incentives Only pro-gram, supplements were provided to families for any amount of work and were more generousfor part-time work than for full-time work. In SSP and New Hope, in contrast, parents had towork full time (at least 30 hours per week) in order to receive the supplement. In addition;whereas MFIP Incentives Only provided the supplement within the welfare system by raising theearnings disregard, SSP and New Hope did so outside the welfare system. Finally, the three pro-grams differed with respect to whether they provided in-kind benefits in addition to cash assis-tance. New Hope earmarked part of the benefits for particular uses (child care and health insur-
40)
-19-
ance), while MFIP Incentives Only and SSP provided a cash supplement that families couldspend as they wished.
The findings presented are for the child samples in the SSP and New Hope evaluations(children of single-parent welfare recipients in SSP4 and children of low-income single parents inNew Hope) and for the children of the urban, single-parent long-term welfare recipients in theMFIP Incentives Only program,5 which included the largest sample of children examined in theMFIP evaluation. Later in this chapter, we present the programs' impacts for children of differentages and for girls and boys. Box 2.1 provides a guide to reading the figures in this monograph.
Figure 2.1 shows the impacts of the three earnings supplement programs that did not pro-vide mandatory employment services on children's achievement as measured by parents' ratingsand children's math test scores. As explained in Box 2.1, each bar represents on a commonscale across programs the difference between the program and control groups. In one of thestudies, only ratings by parents were collected. In the other two programs, additional informationabout the effects of these programs on children's school achievement was obtained from testscores or ratings by teachers.
All the programs that provided earnings supplements without mandatoryemployment services improved children's school achievement.
As can be seen in Figure 2.1, these three earnings supplement programs improved chil-dren's achievement in school, increasing it by about 10 percent to 15 percent of the averagevariation in the control group. These results suggest that these kinds of programs have small butstatistically significant positive effects (for a discussion of the size of these effects, see Box 2.2).The study of the MFIP Incentives Only program, which measured children's achievement solelyon the basis of parents' ratings, found positive effects on this measure. The SSP study, which in-cluded ratings by parents as well as children's test scores, found improvements on both meas-ures. The New Hope study included ratings by both parents and teachers. While New Hope wasfound to have no effects on parents' ratings of children's achievement, the program had a posi-tive impact on teachers' ratings. Interestingly, despite the differences in these programs' ap-proaches for instance, with respect to whether they provided the supplement within or outsidethe welfare system, made the program contingent on full-time employment, or earmarked thesupplements for particular work or family supports all three programs had similar, positiveeffects on children's achievement outcomes.
Programs with earnings supplements had either neutral or favorable ef-fects on children's behavior problems and positive behavior.
Figure 2.2 shows the effects of the same three programs on children's externalizing be-havior problems (as opposed to internalizing prciblems such as depression and anxiety). Separatebars show the effects as measured by parents' and teachers' ratings. While there was a reductionin children's behavior problems in the MFIP Incentives Only program, in SSP and New Hope
4In SSP, the study sample included only welfare recipients who had been receiving welfare for at least one year.5At the time of their random assignment to the program or control group, all the long-term welfare recipients in
the MFIP Incentives Only program had been receiving welfare for at least two years.
-20- 3 7
How Welfare and Work Policies Affect Children
Box 2.1
How to Read the Figures in This Monograph
Most of the figures in this report show the impacts of specific welfare reform initiatives. Below is a tablethat uses data from the SSP evaluation to illustrate the types of information the figures represent. The ta-ble shows for the program group and the control group separately the average rating by parents ofchildren's school achievement (expressed on a five-point scale where 1 means "not very well at all" and 5means "very well") averaged across three academic subjects and children's average score on a math skillstest (expressed as the proportion of items answered correctly).
Cognitive Outcomes, Impacts, and Effect Sizes in the SSP Evaluation
Program Group Control Group Difference Effect SizeAverage Average (Impact)
Parents' ratings of achievement 3.71 3.61 0.10** 0.11**Children's math test scores 0.56 0.52 0.04** 0.14**
The table also shows the differences between the program and control group averages that is, the pro-gram's impacts. The difference between the average achievement ratings given by parents in the programand control groups was .10. The average math test scores of children in the two groups differed by .04; thatis, children in the program group had an average score that was 4 percentage points higher than the averagescore of children in the control group.
The final column of the table converts each impact into an effect size by dividing the difference between theprogram and control group averages on each achievement measure by the standard deviation of the controlgroup outcomes that is, their degree of "spread," or how variable they are relative to the average outcome.This procedure standardizes the impact estimates to allow comparison across measures. In this case, it showsthat the impact of .10 on parents' achievement ratings is similar in magnitude to the impact of .04 on math testscores. Both effect sizes represent a change of about 10 percent to 15 percent of the average variation in theoutcomes. The bar charts used in this report graph these effect sizes as in the figure below. The program andcontrol group averages underlying all the effect sizes shown in the figures are provided in the Appendix.
0.12
0.08
0.04
0.00
0.14**
0.11**
Parents' ratings of Children's math testachievem ent scores
In most of the figures, stars appear above some of the bars, which means that the differences between theprogram and control groups on the measures indicated are statistically significant, that is, unlikely to be dueto chance. In the figures showing impacts for subgroups such as boys and girls, the figure notes indicatewhether the differences between the subgroup impacts are statistically significant. Unless otherwise noted,all the impacts (sometimes called effects) discussed in the text are statistically significant.
-21-
How Welfare and Work Policies Affect Children
Figure 2.1The Three Earnings Supplement Programs Without Mandatory Employment Services
Improved Children's School Achievement
°Parents' reports Teachers' reports or children's test scores
0.45
0.40
0.35
0.30
0.25
0.200.14 *
0.150.11 **
0.10
0.05
0.00
0.14 **
0.08
0.25 **
Enhanced earnings disregard Cash supplement for full-time Cash supplement and other(MFIP Incentives Only) work (SSP) subsidies for full-time work
(New Hope)
NOTES: In each study, children were selected for inclusion in the sample on the basis of their age at random assignment ortheir age at follow-up.
The MFIP sample includes children of parents in the MFIP evaluation aged 5-12 at the time of the three-year follow-up survey (aged approximately 2-9 at the time of random assignment) whose parents were long-term recipients.in urbancounties and underwent random assignment between April 1, 1994, and OctObei 31, 1994 (sample size for MFIP IncentivesOnly = 573).
The SSP sample includes children of single parents in the SSP evaluation aged 6-11 at the time of the three-yearfollow-up survey (aged approximately 3-8 at random assignment) who were living in the home at the time of randomassignment and at the time of the three-year follow-up survey (sample size = 2,158).
The New Hope sample includei children of the single parents in the New Hope evaluation who were aged 1-10 atrandom assignment and whose parents participated in the two-year follow-up survey (sample size = 832).
Statistical significance levels are indicated as: * = 10 percent; ** = 5 percent; *** = 1 percent (two-tailed test).In MFIP, achievement was assessed using a single-item measure that asked parents to rate their child's overall
performance in school on a scale ranging from 1 ("not well at all") to 5 ("very Well").In SSP, achievement was measured using a 26- to 34-item math skills test and expressed in terms of the proportion of
items answered correctly. Parents' assessments of achievement were measured using their ratings of their child's functioning inthree academic subjects on a five-point scale ranging from 1 ("not very well") to 5 ("very well"). The ratings were averagedacross the three academic subjects to compute a single score for each child.
In New Hope, teachers' reports of achievement were measured using the 10-item Academic Subscale from the SocialSkills Rating System, which asked teachers to rate the child's skills relative to those of other children in areas such as math,reading, and oral communication on a five-point scale ranging from 1 ("bottom 10 percent") to 5 ("top 10 percent"). Theresponses were averaged across the 10 items to compute a single score for each child. Parents' assessments of achievement weremeasured using a single-item measure that asked parents to rate their child's school performance, based on past report cards orother sources, on a five-point scale ranging from 1 ("not at all well") to 5 ("very well").
-22-
How Welfare and Work Policies Affect Children
Figure 2.2One Earnings Supplement Program Without Mandatory Employment Services
Reduced Children's Behavior Problems
0 Parents' reports Teachers' reports
0.00
-0.05-0.01 -o.o i
-0.10-0.09
-0.15-0.15 *
-0.20
-0.25
-0.30
-0.35
-0.40
-0.45
Enhanced earnings disregard Cash supplement for full-time Cash supplement and other(MFIP Incentives Only) work (SSP) subsidies for full-time work
(New Hope)
NOTES: In each study, children were selected for inclusion in the sample on the basis of their age at random assignment or
their age at follow-up.The MFIP sample includes children of parents in the MFIP evaluation aged 5-12 at the time of the three-year follow-
up survey (aged approximately 2-9 at the time of random assignment) whose parents were long-term recipients in urban
counties and underwent random assignment between April 1, 1994, and October 31, 1994 (sample size for MFIP Incentives
Only = 573).The SSP sample includes children of single parents in the SSP evaluation aged 6-11 at the time of the three-year
follow-up survey (aged approximately 3-8 at random assignment) who were living in the home at the time of randomassignment and at the time of the three-year follow-up survey (sample size = 2,158).
The New Hope sample includes children of the single parents in the New Hope evaluation who were aged 1-10 atrandom assignment and whose parents participated in the two-year follow-up survey (sample size = 832).
Statistical significance levels are indicated as: * = 10 percent; ** = 5 percent; *** = 1 percent (two-tailed test).In MFIP, behavior problems were measured using parents' responses to a 12-item externalizing subscale of the
Behavioral Problems Index that assesses aggressive behaviors such as bullying and cheating. Responses range from 0 ("nottrue") to 2 ("very true"). The responses to the 12 questions were summed to compute a single score for each child.
In SSP, behavior problems were measured using a four-item externalizing subscale that asked parents to assess theirchild's acting out and aggressive behaviors on a three-point scale ranging from I ("never") to 3 ("often"). The responses wereaveraged across the four items to compute a single score for each child.
In New Hope, behavior problems were measured using a six-item externalizing subscale of the Problem BehaviorScale from the Social Skills Rating System that asked parents and teachers about the child's aggressive behavior and how oftenthe child needed to be disciplined for misbehavior on a five-point scale ranging from 1 ("never") to 5 ("all the time"). Theresponses were averaged across the six items to compute a single score for each child.
-23-4 0
there were no positive or negative effects.6 These findings suggest that, if these programs influ-ence children's behavior problems at all, their effects are favorable.
Turning to children's positive behavior (positive social interactions with others), we seesimilar results, with increases in children's positive behavior in two of the three studies (MFIPIncentives Only and New Hope; see Figure 2.3). Whereas MFIP's positive effects were found forparents' ratings of children's positive behavior (no ratings by teachers were collected), NewHope's were limited to teachers' ratings (although ratings by both parents and teachers were col-lected). As with behavior problems, favorable effects on positive behavior were observed insome but not all studies, but the effects found were never unfavorable.
Programs with earnings supplements had either neutral or positive effectson children's health.
Only two of the studies, MFIP Incentives Only and SSP, included data on children'shealth, and in both cases these data were parents' ratings. SSP increased ratings of children'shealth (see Figure 2.4), but this finding was not replicated in MFIP Incentives Only. Thus, aswith the findings on children's behavioral outcomes, the single effect that was observed was fa-vorable but was not found in MFIP Incentives Only.
How Welfare and Work Policies Affect ChildrenBox 2.2
How Large Are These Effect Sizes?
Understanding effect sizes
One way to understand how large the effects of the programs considered here are is tocompare them with the programs' effects on related measures. For example, Full MFIP had an ef-fect on school achievement of .15 and an effect on behavior problems of .17. The program alsoreduced the proportion of children performing below average in school by 5 percentage points andthe proportion of children with a high level of behavior problems by 8 percentage points.'
Another way of judging whether a given effect is large or small is to relate its size to thekinds of percentile scores associated with school tests. Children in the studies examined in thismonograph scored at around the 25th percentile on most of the standardized tests they took (mean-ing that 25 percent of children nationally scored below the average child in the control group in thestudies examined here). A positive impact on test scores with an effect size of .15 would amount toabout a 5 percentile point improvement, implying that their average percentile ranking rose fromthe 25th to the 30th percentile (so that at the end of the follow-up period they would rank above 30percent of children nationally). An effect size of .25 amounts to about an 8 percentile point differ-ence, implying that their average percentile ranking rose to the 33`d percentile.
Similar logic applies to the effect sizes on all the measures reported. Suppose, for example,that a program has an effect of size .15 on children's positive behavior. This would imply a child atthe 25th percentile of the positive behavior distribution who experienced the average program im-pact would be at the 30th percentile of the positive behavior distribution at the end of the study.
(continued)
6As reported later in this chapter, New Hope decreased boys' behavior problems but increased girls' behaviorproblems.
-24- 41
Box 2.2 (continued)
How do these effects compare with those on adult economic outcomes?
The effects on children of the earnings supplement programs are about one-half to one-third as large as those on adult economic outcomes. Generally, the programs increased employ-ment rates by about 10 percentage points and reduced poverty by a similar amount. Most of theimpacts correspond to effect sizes of about .30, but on selected measures of employment they areas large as .50. The sizes of the effects on children are generally around .15.
How do these effects compare with those of other interventions aimed directly or indirectlyat children?
Other programs aimed at improving outcomes for both parents and children specifi-cally, parental employment and children's development include home-visiting programs, inwhich nurses or other paraprofessionals provide one-on-one case management to link low-incomeparents with services in their communities. Some such programs combine home-visiting servicesfor parents with child development programs for children. A recent review, however, suggests thathome-visiting programs are not always effective in improving parents' and children's function-ing.2 However, studies of home-visiting programs in which improvements in children's well-beingwere observed found effect sizes as large or larger than those presented here. For example, studiesof the Infant Health and Development Program, a large-scale two-generational program combin-ing home visits with child development centers and peer support groups for parents of low birth-weight infants, reported effects ranging in size from about .40 to .80 on standardized test scoresand an effect of -.18 on children's behavior problems.3
Turning to programs directly targeted at children, the best known preschool programshave effect sizes much larger than those presented here and can be as large as 1.0.4 Such large ef-fect sizes are not surprising considering that these are the most successful of childhood interven-tions and that they target children directly rather than being designed to affect children indirectlythrough changes in parents' circumstances and behavior. Interestingly, however, the effects ofpreschool programs on test scores and school achievement are much more consistent than their ef-fects on behavioral outcomes, which some studies have found to be neutral or even negative.5 In aset of well-known experiments that investigated the effects of class size on children's school per-formance, children were randomly assigned to classes of different sizes. Classes of smaller sizeimproved children's math and reading grades, with effect sizes ranging from approximately .15 to.35.6 The effects of this intervention, which was targeted directly at school-aged children, arecomparable in size to those of the parent-targeted programs examined here.
Are these effects predictive of children's long-term outcomes?
In assessing the importance of a program's effects on children, we should consider notonly their size but also their relation to later outcomes. Longitudinal studies have found that chil-dren's achievement and behavior problems can have important implications for their well-being inadolescence and adulthood.' Moreover, small differences between children in school achievementearly on can translate into larger differences later.' Therefore, a program's effects on children,even if the effects are small, may continue to have implications over the course of their lives.
'Gennetian and Miller, 2000.2Gomby, Culross, and Behrman, 1999.3lnfant Health Development and Program (IHDP), 1990.4Shonkoff and Phillips, 2000.5Yoshikawa, 1995.6Word, Johnston, Bain, Fulton, Zaharias, Achilles, Lintz, Folger, and Breda, 1990; Krueger, 1997.'Carpi, Wright, Moffit, and Silva, 1998; Masten, Coatsworth, Neemann, Gest, Tellegen, and Garmezy, 1995.sEntwistle, 1985.
4 2-25-
How Welfare and Work Policies Affect Children
Figure 2.3Two Earnings Supplement Programs Increased
Children's Positive Behavior
°Parents' reports Teachers' reports
06 0.40
0.35
0.30
**8 0.25 0.21
0.20 0.18**
o. 0.15E
0.100
g 0.05
ti 0.00
-0.05 -0.02 -0.01
Enhanced earnings disregard Cash supplement for full-time Cash supplement and other(MFIP Incentives Only) work (SSP) subsidies for full-time work
(New Hope)
NOTES: In each study, children were selected for inclusion in the sample on the basis of their age at random assignmentortheir age at follow-up.
The MFIP sample includes children of parents in the MFIP evaluation aged 5-12 at the time of the three-year follow-up survey (aged approximately 2-9 at the time of random assignment) whose parents were long-term recipients in urbancounties and underwent random assignment between April 1, 1994, and October 31, 1994 (sample size for MFIP IncentivesOnly-= 573).
The SSP sample includes children of single parents in the SSP evaluation aged 6-11 at the time of the three-yearfollow-up survey (aged approximately 3-8 at random assignment) who were living in the home at the time of randomassignment and at the time of the three-year follow-up survey (sample size = 2,158).
The New Hope sample includes children of the single parents in the New Hope evaluation who were aged 1-10 atrandom assignment and whose parents participated in the two-year follow-up survey (sample size = 832).
Statistical significance levels are indicated as: * 10 percent; ** = 5 percent; *** = 1 percent (two-tailed test).In MFIP, positive behavior was measured with the 25-item Positive Behavior Scale, which included three subscales:
compliance, social competence, and autonomy. Parents responded to each item on an 11-point scale ranging from 0 ("not at alllike my child") to 10 ("completely like my child"). The responses to the 25 questionswere summed to compute a single scorefor each child.
In SSP, positive behavior was measured using the five-item Positive Social Behavior subscale, which asked parentsto assess their child's prosocial interactions with peers on scale ranging from 1 ("never") to 3 ("often"). The responses wereaveraged across the five items to compute a single score for each child.
In New'Hope, the child's positive behavior was measured using the 25-item Positive Behavior Scale, which includedthree subscales: compliance, social competence, and autonomy. Parents and teachers responded to each item on a five-pointscale ranging from 1 ("never") to 5 ("all of the time"). The responses were averaged across the 25 items to compute a singlescore for each child.
4 3
-26-
How Welfare and Work Policies Affect Children
Figure 2.4One Earnings Supplement Program Improved
Children's Health as Reported by Parents
0.45
0.40
0.35
0.30
0.25
0.20
0.15**
0.11
0.10 0.06
0.05
0.00
Enhanced earnings disregard Cash supplement for full-time work(MFIP Incentives Only) (SSP)
NOTES: In each study, children were selected for inclusion in the sample on the basis of their age at random assignment ortheir age at follow-up.
The MFIP sample includes children of parents in the MFIP evaluation aged 5-12 at the time of the three-year follow-up survey (aged approximately 2-9 at the time of random assignment) whose parents were long-term recipients in urbancounties and underwent random assignment between April 1, 1994, and October 31, 1994 (sample size for MFIP IncentivesOnly = 573).
The SSP sample includes children of single parents in the'SSP evaluation aged 6-11 at the time of the three-yearfollow-up survey (aged approximately 3-8 at random assignment) who were living in the home at the time of randomassignment and at the time of the three-year follow-up survey (sample size = 2,158).
Statistical significance levels are indicated as: * = 10 percent; ** = 5 percent; *** = 1 percent (two-tailed test).In MFIP, health was assessed using a single-item measure that asked parents to rate their child's health on a five-point
scale ranging from 1 ("poor") to 5 ("very good").In SSP, health was measured using parents' responses to four questions about their child's health on a scale ranging
from 1 ("false") to 5 ("true"). The responses were averaged across the four items to compute a single score for each child.
4 4
-27-
Overall; the programs that provided earnings supplements without man-datory employment services improved child outcomes. Encouragingly,this finding suggests that such programs can both increase family re-sources and improve child outcomes.
The findings are most consistent with regard to school achievement. Effects on children'sbehavioral and health outcomes were not found across all studies, but those that were observedwere favorable.
These results are consistent with those of the studies reviewed in Chapter 1, most of whichreported positive associations between family income and children's well-being, particularly as re-flected in cognitive performance and school achievement. The fact that welfare programs with anantipoverty component can lead to improvements in children's cognitive outcomes improve-ments that are detected two to three years after their parents first enter the programs has impor-tant implications for policy and program design. Furthermore, the consistency in the findings acrossthe sites and studies considered here justifies greater confidence in the generalizability of the pro-grams' effects. At the same time, most of the effects are small and are concentrated in children'scognitive development rather than their behavior or health. It is not yet clear how to improve chil-dren's behavior and health systematically across programs.
III. Adding Mandatory Employment Services to an Earnings Supplement
A given program feature may work differently when combined with other features thanwhen studied in isolation. Because most state welfare programs are packages of features, under-standing how the features interact is critical to informing welfare policy. In this section, we ex-amine how earnings supplements interact with mandatory employment services.
The MFIP evaluation permits a direct experimental comparison of earnings supplementprograms with and without mandatory employment services. Whereas in Full MFIP parents wereboth subject to a participation mandate and provided with an earnings supplement, in MFIP In-centives Only parents were subject to all the same program features except the mandate. Theevaluation provides a rigorous test of the effects of adding a mandate to supplements becauseparents were randomly assigned to one of these two groups or to a control group that was not of-fered either of the MFIP programs. Random assignment allows us to be confident that the differ-ences between the two programs' effects are caused by the differences between the programsrather than by any unmeasured differences between participants.
Both MFIP Incentives Only and Full MFIP increased employment for long-term welfarerecipients. However, only Full MFIP led to an increase in full-time employment (30 hours ormore per week), while the MFIP Incentives Only program boosted only part-time employment.From this result it appears that adding the mandate increased full-time employment relative toproviding only earnings supplements. How do these increases in employment affect child out-comes in the two programs? One might hypothesize that higher employment has negative effectson children that undermine the positive effects of the higher family income resulting from earn-ings supplements. Alternatively, one might hypothesize that increases in full-time employmentbenefit children by making family routines more regular and providing children with positive rolemodels. Another hypothesis is that mandatory employment services increase parental stress, irre-
-28- 4 5
spective of their effects on employment, and thereby reduce the positive effects of earnings sup-plements on children's well-being:
The impacts of the two MFIP programs on children are presented in Figure 2.5. The find-ings suggest that adding the participation mandate had no effect on parents' ratings of children'sachievement in school, behavior problems, or health. The only outcome that appears to have beenaffected by the addition of the mandate was parents' ratings of children's positive behavior:While the effect is positive in the MFIP Incentives Only program, there is no effect in the FullMFIP program. Notably, Full MFIP did not produce any negative effects on children, even in thecase of positive behavior.
Although drawn from a single study, these findings show that mandatoryemployment services need not reduce the positive effects on children ofprograms with earnings supplements.
Programs with earnings supplements appear to increase children's achievementwhether the supplements are provided alone or combined with a participation mandate that in-creases full-time work.
IV. How Might These Programs Have Affected Children? Effects onChild Care, Parents' Emotional Well-Being, and Parenting Behavior
There are a number of possible explanations for the effects of the earnings supplement pro-grams discussed above. Through what mechanisms might the increases in parental employment andincome caused by the programs benefit children? One possibility is that increased employment andfinancial stability improve parents' emotional well-being or reduce their feelings of stress. This, inturn, may have a favorable influence on parents' interactions with their children. Moreover, by in-creasing the use of child care (because of higher employment) and changing the type of child carethat parents use (because of wider child care options made possible by higher income), earningssupplement programs introduce children to environments and educational opportunities to whichthey otherwise might not have been exposed. This, in turn, may enhance children's emotional andcognitive development. All these are mechanisms through which programs like New Hope, FullMFIP, MFIP Incentives Only, and SSP might have affected children.
It is critical to learn more about the mechanisms through which programs affect childrenand families because programs cannot always be fully replicated in new sites or locales, wheredifferent combinations of program features may be adopted or different factors such as childcare or parenting practices may be specifically targeted. Attributing profgam effects on chil-dren to one or more specific mechanisms requires complicated statistical analyses beyond thescope of this synthesis. As outlined in more detail at the end of the monograph, future documentsfrom the Next Generation project will present such analyses. In the present context, we simplyexplore the pathways by which these programs likely affected children.
Possible mediators of program effects on child outcomes are child care, family relations,parental well-being, and parenting practices. Evidence from the studies reviewed here suggestthat the four programs with generous earnings supplements affected a number of these possiblemediating factors in generally positive ways; however, the nature of the effects varied across
-29- 4 6
0.20
0.15
0.10
0.05
0.00
-0.05
-0.10
-0.15
-0.20
-0.25
How Welfare and Work Policies Affect Children
Figure 2.5Adding Mandatory Employment Services to an Earnings Supplement Program
Affected None of the Effects for Children Except Positive Behavior
0 Enhanced earnings disregard (MFIP Incentives Only)MI Enhanced earnings disregard with work mandate (Full MFIP)
0.18**
0.14 * °A 5
0.01
0.06
-0.17 **
-0.09
-0.15*
School Achievement Behavior Problems Positive Behavior Health
NOTES: The MFIP sample includes children of parents in the MFIP evaluation aged 5-12 at the time of the three-year follow-up survey (aged approximately 2-9 at the time of random assignment) whose parents were long-term recipients in urbancounties and underwent random assignment between April 1, 1994, and October 31, 1994 (sample size for Full MFIP = 587;sample size for MFIP Incentives Only = 573).
The statistical significance levels of the impacts are indicated as: * = 10 percent; ** = 5 percent; *** = 1 percent (two-tailed test). The statistical significance levels of the differences between impacts are not noted in the figure. The only differencebetween impacts that was statistically significant was that in positive behavior for MFIP Incentives Only and Full MFIP.
Achievement was assessed using a single-item measure that asked parents to rate their child's overall performance inschool on a scale ranging from 1 ("not well at all") to 5 ("very well").
Behavior problems were measured using parents' responses to a 12-item externalizing subscale of the BehavioralProblems Index that assesses aggressive behaviors such as bullying and cheating. Responses range from 0 ("not true") to 2("very true"). The responses to the 12 questions were summed to compute a single score for each child.
Positive behavior was measured with the 25-item Positive Behavior Scale, which included three subscales:compliance, social competence, and autonomy. Parents responded to each item on an 11-point scale ranging from 0 ("not at alllike my child") to 10 ("completely like my child"). The responses to the 25 questions were summed to compute a single scorefor each child.
Health was assessed using a single-item measure that asked parents to rate their child's health on a five-point scaleranging from 1 ("poor") to 5 ("very good").
-30-
programs. New Hope increased the use of formal child care for young children, and other re-search has found that formal care arrangements are associated with greater cognitive readinessfor school and improved social functioning among children.7 Full MFIP also increased the use offormal and stable child care for children of long-term welfare recipients, but the MFIP IncentivesOnly program did not. SSP increased children's participation in after-school activities. With re-gard to family relations, both Full MFIP and MFIP Incentives Only increased marriage amongsingle-parent long-term welfare recipients, and marriage may have positive effects on children'swell-being. However, neither SSP nor New Hope produced similar effects on marriage rates.8Single-parent long-term welfare recipients in both the Full MFIP and MFIP Incentives Only pro-grams were also less likely to report experiencing domestic abuse, an outcome that was notmeasured in New Hope or SSP. In some studies, there were also improvements in parents' emo-tional well-being: MFIP Incentives Only reduced parental depression among long-term welfarerecipients, and New Hope reduced parental stress and increased parents' feelings of hope. Butparents in New Hope also reported feeling more time pressure, and, surprisingly, SSP increaseddepression. In terms of parenting behavior, the programs had very few effects on the quality ofparents' interactions with their children, one of the key ways in which increases in employmentand income are expected to affect children.
All these findings illustrate how difficult it is to conclusively attribute the programs' ef-fects on children to one mechanism. None of the outcomes considered to be possible mediatorsof effects on children was affected across all programs, at least according to the measures exam-ined. All four programs have in common one program feature (a generous earnings supplement)and one result (an increase in employment and income), but the way in which these factors mayhave affected children remains unclear.
V. How Did Children and Families in the Program Groups Fare?
Effect size and impact analyses tell us nothing about how the children and families inearnings supplement programs fared only how the program and control groups fared relativeto one another. To get a sense of the absolute level of functioning in the families that experiencedthe programs, in this section we examine program group outcomes rather than effect sizes or im-pacts specifically, how program group members fared according to several indicators of well-being for families, parents, and children.
Table 2.1 presents the outcome levels for single-parent families in the program groupswho, at the time these measures were assessed, were already eligible for the earnings supplementand who had, on average, experienced more favorable outcomes than their counterparts in thecontrol groups. Because some measures were available for more than one study, for each meas-ure a range of outcome levels is presented. As indicated in the table, two-thirds to three-quartersof parents in the program groups had incomes from earnings and benefits that put them below the
7Zaslow, Oldham, Moore, and Magenheim, 1998; Zaslow, McGroder, Moore, and LeMenestrel, 1999.81n one province, SSP increased marriage rates; in the other province, it reduced marriage rates. New Hope had
no effects on marriage rates.
48-31-
How Welfare and Work Policies Affect. hildren
Table 2.1
Program Group Outcomes for Earnings Supplement Programs
Outcome
AverageLevel inProgram
Grouv(s), %
Number ofStudies
Outcome WasMeasured
ProgramImproved
Outcomes on
This Measurea
Material hardship
Parent has income below poverty 69 - 73 2 V
Any child skipped meal due to lack of money for food 5 - 6 2
Health insurance coverage
Child with continuous health insurance coverage over 3 years 76 - 79 2 V
Neighborhood quality
Lives in a safe neighborhood 73 - 76 2
Parents' well-being
Parent at risk of clinical depression 23 -49 3
Parent stressed much or all of the time 49 1
Parent physically abused by partner last year 22 2
Children's well-being
Child repeated grade since random assignment 4 - 14 3
Child received special education since random assignment 14 - 21 3
Child with high behavioral or emotional problems 7 - 11 2
Child at or below 25th percentile on language skills 38 1
Child with long-term health problems 32 1
Child in very good or excellent health 75 - 80 2
NOTES: In each study, children were selected for inclusion in the sample on the basis of their age at random assignment or theirage at follow-up. The programs with earnings supplements are the four programs in the MFIP, SSP, and New Hope evaluations.
The MFIP sample includes children (of parents in the MFIP evaluation) aged 5-12 at the time of the three-year follow-up survey (aged approximately 2-9 at the time of random assignment) whose parents were long-term recipients in urban countiesand underwent random assignment between April 1, 1994, and October 31, 1994 (sample size for Full MFIP = 587; for MFIPIncentives Only = 573).
The SSP sample includes children (of parents in the SSP evaluation) aged 6-11 at the time of the three-year survey(aged approximately 3-8 at random assignment) who were living in the home at the time of random assignment and at the time ofthe three-year follow-up survey (sample size = 2,158).
The New Hope sample includes children aged 1-10 at random assignment of the single-parent members of the NewHope evaluation who participated in the two-year follow-up survey (sample size = 832).
aA check mark signifies that the outcomes on the measure were improved in at least one study.
-32-4 9
poverty line during the study period, indicating that their incomes were generally very low de-spite the increases in income produced by the programs. A very small proportion of programgroup families seemed to be living in extreme deprivation, with 5 percent to 6 percent of parentsreporting that their children were forced to skip meals because of lack of money for food. Three-quarters of parents had children who had continuous health insurance coverage over the previousthree years, and three-quarters of parents reported living in what they deemed to be safeneighborhoods.
The findings also suggest that these single parents the great majority of whom weremothers and their children were still at risk of psychological, physical, and cognitive problemsafter being in the programs. According to self-reports collected from parents two or three yearsafter random assignment, one-quarter to one-half of parents were at risk of clinical depression,one-half felt stressed much or all of the time, and one-fifth had been physically abused by a part-ner in the previous year. As for their children, since random assignment around 10 percent hadrepeated a grade, about one-fifth had received special education, and about 10 percent had a highlevel of behavioral or emotional problems. More than one-third of children scored at or below the25th percentile on a nationally standardized test of language skills. A comparable proportion ofchildren had long-term health problems, although most parents reported that their children werein very good or excellent health.
Despite the benefits to families of the earnings supplements, many chil-dren and families in such programs remained at risk.
Even after improvements resulting from the programs are taken into account, many chil-dren with parents in the earnings supplement programs were still struggling, as were their par-ents. While the positive effects on children were measurable and consistent across studies, theyclearly do not obviate the need for other interventions targeted at low-income children.
VI. Effects on Long-Term Welfare Recipients
Much of the impetus behind welfare reform arose from concern about long-term welfarerecipients, families who were using welfare not as a temporary safety net but for long periods oftime. In the current welfare climate, parents in these families face the greatest barriers to em-ployment because they are least likely to have the skills to move from welfare to employment.They are also most likely to be affected by welfare time limits. As caseloads decline, these fami-lies are most likely to remain on the rolls.
The positive effects of earnings supplements on employment and incomeare larger for long-term welfare recipients than for recipients with ashorter welfare history.
Short-term welfare recipients are more likely than long-term recipients both to have a jobhistory and work skills and to work in the absence of a welfare-to-work program. Therefore, thedifferences between program and control groups are generally smaller among short-term recipi-
5 0
-33-
ents than among their long-term counterparts in programs like MFlP.9'1° As discussed in detailbelow, the average increase in income is also larger among long-term welfare recipients; it istherefore likely that effects on children are most pronounced in these families as well.
In the figures presented earlier in this chapter, data on children in the MFIP IncentivesOnly and Full MFIP programs are presented for families who had received welfare for two yearsor more. In the Appendix, we compare the effects of the two MFIP programs with those of SSP"and New Hope for this same subgroup of families to see whether the cross-program consistencyof the findings in the child samples examined earlier held in the long-term recipient subgroups(see Appendix Table 2). For this subgroup, both Full MFIP and MFIP Incentives Only raisedparents' ratings of children's school achievement. Similarly, New Hope increased both parents'and teachers' ratings of children's achievement for long-term recipients. (Note that these NewHope effects are even larger than those reported above for the child sample examined earlier,which included the children of all low-income single parents.) SSP had positive effects on par-ents' ratings of children's school achievement for parents with two or more years of welfare re-ceipt, as the program did for families with a year or more of welfare receipt (as presented earlierin this monograph).
There is a similar pattern of positive effects for children's behavior problems, althoughhere the effects are smaller and less consistent across studies (see Appendix Table 2). Reductionsin behavior problems were found among long-term welfare recipients, but only for Full MFIPand MFIP Incentives Only. The patterns in New Hope are similar but not statistically significant,and no such effects were found in SSP. A similar pattern of effects, with two programs (in thiscase, MFIP Incentives Only and New Hope) producing favorable effects, was observed for chil-dren's positive behavior.
While the programs were more successful in improving outcomes for children of long-term than short-term recipients, children in control group families with long-term welfare receiptfared worse than children in control group families with a shorter welfare history (often calledrecent applicants). For example, on average, children in recent applicant families were ratedhigher on positive behavior and lower on behavior problems.12 Children in recent applicant fami-lies were also about half as likely to be performing below average in school and about half aslikely to be suspended or expelled as their peers in long-term recipient families. Given that par-ents in long-term recipient families face greater employment barriers and have fewer job skills, itis not surprising that their children also face greater difficulties.
9Berlin, 2000; Knox et al., 2000.'Notably, not all earnings supplement programs are less effective at increasing employment and income among
people with a shorter history of welfare receipt. For example, a companion to the SSP study examined in this mono-graph (for which data on children have not yet been analyzed) tested the effect of offering the SSP earnings supple-ment to welfare applicants, but only after they had been on welfare for at least one year during the study period. Theone-year waiting period boosted the program's effectiveness by reducing the number of applicants receiving the sup-plement who would have started working even without the supplement (Berlin, 2000).
"The SSP sample consisted of single parents who had been on welfare during 11 of the previous 12 months. Inthis analysis, single parents who had been on welfare for at least two years were treated as long-term welfare recipi-ents to make the sample more comparable to that in the MFIP study.
12Gennetian and Miller, 2000.
-34-
The positive impacts of programs with earnings supplements on childrenare particularly pronounced among long-term welfare recipients.
In part, the larger child impacts in families with long-term welfare receipt may be a resultof the programs' larger effects on long-term recipients' income and employment. As a result ofthese positive impacts, program group children in this subgroup perform at a level near that ofcontrol group children in families that had received welfare for less than two years. In otherwords, the earnings supplement programs increased these children's average outcomes to thelevel of the highest-functioning children in these low-income samples.
VII. Differences Between Preschool-Aged and Early School-Aged Children
In light of the research on the effects of child care and income on children reviewed ear-lier in this chapter, which indicated that the association between family income and schoolachievement is stronger for younger than for older children and that child care environments haveimportant effects on younger children's functioning, one might expect the impact of earningssupplement programs on achievement to be stronger and more positive for preschool-aged chil-dren than for children in middle childhood, although it is less clear what the age differences inimpacts on social behavior, if any, might be. In this section, we examine the effects of earningssupplement programs separately on the basis of children's age when parents first enrolled in thestudy, comparing impacts for children who were 3-5 years old with those for children who were6-9 years old. Assessments of the children were collected two to three years after parents' ran-dom assignment.
In general, the programs that included earnings supplements had positive impacts on chil-dren in both age groups; however, some interesting patterns of effects on children's social behav-ior were observed. The impacts on achievement did not consistently favor younger or older chil-dren (see Figure 2.6). The impacts of SSP were larger for younger than for older children. Thepattern of impacts in the New Hope study depended on whether parents or teachers assessed thechild. Finally, the impacts of the two MFIP programs were stronger for the older than theyounger children. However, none of these subgroup differences were statistically significant.
The programs' effects on social behavior show more pronounced age differences, butthese differences were not statistically significant (see Figure 2.7). The impacts were generallylarger for older than for younger children, with the older children more often showing positiveimpacts on social behavior (reductions in behavior problems). This finding is reinforced by ouranalyses of positive as well as negative aspects of children's behavior (not shown in figures).Again, none of these differences for younger and older children were statistically significant.
VIII. Differences Between Boys and Girls
Separate analyses of the effects of earnings supplement programs on girls' and boys'school achievement revealed no large gender differences (see Figure 2.8). Only in the case ofMFIP Incentives Only did the impacts differ by gender, and those impacts were more pronouncedfor girls than for boys. In SSP, the impacts were positive for both boys and girls. The impacts onchildren's test scores in New Hope and on parents' ratings of achievement in both New Hope and
-35-
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
0.23
0.07
How
Wel
fare
and
Wor
k Po
licie
s A
ffec
t Chi
ldre
n
Figu
re 2
.6T
he E
ffec
ts o
n Sc
hool
Ach
ieve
men
t Did
Not
Dif
fer
Con
sist
ently
Acr
oss
Ear
ning
s Su
pple
men
t Pro
gram
s by
Chi
ldre
n's
Age
at R
ando
m A
ssig
nmen
t
°Par
ents
' rep
orts
Tea
cher
s' r
epor
ts o
r ch
ildre
n's
test
sco
res
0.22
**
0.09
0.05
0.31
**
0.20
0.15
0.30
**
Age
d 3-
5A
ged
6-9
Enh
ance
d ea
rnin
gs d
isre
gard
(MFI
P In
cent
ives
Onl
y)
Age
d 3-
5A
ged
6-9
Age
d 3-
5A
ged
6-9
Age
d 3-
5A
ged
6-9
Cas
h su
pple
men
t for
ful
l-tim
eC
ash
supp
lem
ent a
nd o
ther
sub
sidi
es f
orE
nhan
ced
earn
ings
dis
rega
rd w
ithw
ork
(SSP
)fu
ll-tim
e w
ork
(New
Hop
e)w
ork
man
date
(Fu
ll M
FIP)
NO
TE
S: I
n ea
ch s
tudy
, chi
ldre
n w
ere
sele
cted
for
incl
usio
n in
the
sam
ple
on th
e ba
sis
of th
eir
age
at r
ando
m a
ssig
nmen
t or
thei
rag
e at
fol
low
-up.
The
MFI
P sa
mpl
e in
clud
es c
hild
ren
of p
aren
ts in
the
MFI
P ev
alua
tion
aged
6-1
2 at
the
time
of th
e th
ree-
year
fol
low
-up
surv
ey (
aged
app
roxi
mat
ely
3-9
atth
e tim
e of
rand
om a
ssig
nmen
t) w
hose
par
ents
wer
e lo
ng-t
erm
rec
ipie
nts
in u
rban
cou
ntie
s an
d un
derw
ent r
ando
m a
ssig
nmen
t bet
wee
n A
pril
1, 1
994,
and
Oct
ober
31,
199
4 (s
ampl
e si
ze f
orFu
ll M
FIP
= 4
88; s
ampl
e si
ze f
or M
FIP
Ince
ntiv
es O
nly
= 4
72).
The
SSP
sam
ple
incl
udes
chi
ldre
n of
sin
gle
pare
nts
in th
e SS
P ev
alua
tion
aged
6-1
1 at
the
time
of th
e th
ree-
year
fol
low
-up
surv
ey (
aged
app
roxi
mat
ely
3-8
at r
ando
mas
sign
men
t) w
ho w
ere
livin
g in
the
hom
e at
the
time
of r
ando
m a
ssig
nmen
t and
at t
he ti
me
of th
e th
ree-
year
fol
low
-up
surv
ey (
sam
ple
size
=2,
158)
.T
he N
ew H
ope
sam
ple
incl
udes
chi
ldre
n of
the
sing
le p
aren
ts in
the
New
Hop
e ev
alua
tion
who
wer
e ag
ed 3
-9 a
t ran
dom
ass
ignm
ent a
nd w
hose
par
ents
part
icip
ated
in th
etw
o-ye
ar f
ollo
w-u
p su
rvey
(sa
mpl
e si
ze =
546
).T
he s
tatis
tical
sig
nifi
canc
e le
vels
of
the
impa
cts
are
indi
cate
d as
: * =
10
perc
ent;
** =
5 p
erce
nt; *
** =
1 p
erce
nt (
two-
taile
d te
st).
The
stat
istic
al s
igni
fica
nce
leve
ls o
f th
edi
ffer
ence
s be
twee
n im
pact
s ar
e no
t not
ed in
the
figu
re. N
one
of th
e di
ffer
ence
s be
twee
n im
pact
s fo
r th
e ch
ildre
n ag
ed 3
-5 a
nd th
e ch
ildre
n ag
ed 6
-9w
ere
stat
istic
ally
sig
nifi
cant
.In
MFI
P, a
chie
vem
ent w
as a
sses
sed
usin
g a
sing
le-i
tem
mea
sure
that
ask
ed p
aren
ts to
rat
e th
eir
child
's o
vera
ll pe
rfor
man
ce in
sch
ool o
n a
scal
e ra
ngin
g fr
om1
("no
t wel
lat
all"
) to
5 (
"ver
y w
ell"
).In
SSP
, ach
ieve
men
t was
mea
sure
d us
ing
a 26
- to
34-
item
mat
h sk
ills
test
and
exp
ress
ed in
term
s of
the
prop
ortio
n of
item
s an
swer
ed c
orre
ctly
. Par
ents
'as
sess
men
ts o
fac
hiev
emen
t wer
e m
easu
red
usin
g th
eir
ratin
gs o
f th
eir
child
's f
unct
ioni
ng in
thre
e ac
adem
ic s
ubje
cts
on a
fiv
e-po
int s
cale
ran
ging
fro
m I
("n
otve
ry w
ell"
) to
5 (
"ver
y w
ell"
). T
hera
tings
wer
e av
erag
ed a
cros
s th
e th
ree
acad
emic
sub
ject
s to
com
pute
a s
ingl
e sc
ore
for
each
chi
ld.
In N
ew H
ope,
teac
hers
' rep
orts
of
achi
evem
ent w
ere
mea
sure
d us
ing
the
10-i
tem
Aca
dem
ic S
ubsc
ale
from
the
Soci
al S
kills
Rat
ing
Syst
em, w
hich
ask
ed te
ache
rs to
rat
eth
e ch
ild's
ski
lls r
elat
ive
to th
ose
of o
ther
chi
ldre
n in
are
as s
uch
as m
ath,
rea
ding
, and
ora
l com
mun
icat
ion
on a
fiv
e-po
int s
cale
ran
ging
fro
m 1
("bo
ttom
10
perc
ent"
) to
5 (
"top
10
perc
ent"
). T
he r
espo
nses
wer
e av
erag
ed a
cros
s th
e 10
item
s to
com
pute
a s
ingl
e sc
ore
for
each
chi
ld. P
aren
ts' a
sses
smen
ts o
f ac
hiev
emen
t wer
e m
easu
red
usin
ga
sing
le-i
tem
mea
sure
that
ask
ed p
aren
ts to
rat
e th
eir
child
's s
choo
l per
form
ance
, bas
ed o
n pa
st r
epor
t car
ds o
r ot
her
sour
ces,
on
a fi
ve-p
oint
sca
le r
angi
ng f
rom
I("
not a
t all
wel
l")
to 5
("v
ery
wel
l").
0.05
0.00
-0.0
5
-0.1
0
-0.1
5
-0.2
0
-0.2
5
-0.3
0
-0.3
5
-0.4
0
How
Wel
fare
and
Wor
k Po
licie
s A
ffec
t Chi
ldre
n
Figu
re 2
.7E
arni
ngs
Supp
lem
ent P
rogr
ams
Dec
reas
ed B
ehav
ior
Prob
lem
s So
mew
hat M
ore
for
Chi
ldre
nW
ho W
ere
Old
er T
han
for
Chi
ldre
n W
ho W
ere
You
nger
at R
ando
m A
ssig
nmen
t
°Par
ents
' rep
orts
Tea
cher
s' r
epor
ts
0.01
0.01
0.02
IIt
I
-0.0
1-0
.01
-0.1
9
-0.3
2
-0.0
2-0
.05
*
Age
d 3-
5
-0.3
8 *
Age
d 6-
9
Enh
ance
d ea
rnin
gs d
isre
gard
(MF
IP In
cent
ives
Onl
y)
Age
d 3-
5A
ged
6-9
Cas
h su
pple
men
t for
full-
time
wor
k (S
SP
)
Age
d 3-
5.
Age
d 6-
9A
ged
3-5
Age
d 6-
9
Cas
h su
pple
men
t and
oth
er s
ubsi
dies
for
Enh
ance
d ea
rnin
gs d
isre
gard
with
full-
time
wor
k (N
ew H
ope)
wor
k m
anda
te (
Ful
l MF
IP)
NO
TE
S: I
n ea
ch s
tudy
, chi
ldre
n w
ere
sele
cted
for
incl
usio
n in
the
sam
ple
on th
e ba
sis
of th
eir
age
at r
ando
m a
ssig
nmen
t or
thei
r ag
e at
follo
w-u
p.T
he M
FIP
sam
ple
incl
udes
chi
ldre
n of
par
ents
in th
e M
FIP
eval
uatio
n ag
ed 6
-12
at th
e tim
e of
the
thre
e-ye
ar f
ollo
w-u
p su
rvey
(ag
ed a
ppro
xim
atel
y 3-
9 at
the
time
ofra
ndom
ass
ignm
ent)
who
se p
aren
ts w
ere
long
-ter
m r
ecip
ient
s in
urb
an c
ount
ies
and
unde
rwen
t ran
dom
ass
ignm
ent b
etw
een
Apr
il 1,
199
4, a
nd O
ctob
er 3
1, 1
994
(sam
ple
size
for
Full
MFI
P =
488
; sam
ple
size
for
MFI
P In
cent
ives
Onl
y =
472
).T
he S
SP s
ampl
e in
clud
es c
hild
ren
of s
ingl
e pa
rent
s in
the
SSP
eval
uatio
n ag
ed 6
-11
at th
e tim
e of
the
thre
e-ye
ar f
ollo
w-u
p su
rvey
(ag
ed a
ppro
xim
atel
y 3-
8at r
ando
mas
sign
men
t) w
ho w
ere
livin
g in
the
hom
e at
the
time
of r
ando
m a
ssig
nmen
t and
at t
he ti
me
of th
e th
ree-
year
fol
low
-up
surv
ey (
sam
ple
size
=2,
158)
.T
he N
ew H
ope
sam
ple
incl
udes
chi
ldre
n of
the
sing
le p
aren
ts in
the
New
Hop
e ev
alua
tion
who
wer
e ag
ed 3
-9 a
t ran
dom
ass
ignm
ent a
nd w
hose
par
ents
part
icip
ated
in th
e
two-
year
fol
low
-up
surv
ey (
sam
ple
size
= 5
46).
The
sta
tistic
al s
igni
fica
nce
leve
ls o
f th
e im
pact
s ar
e in
dica
ted
as: *
= 1
0 pe
rcen
t; **
= 5
per
cent
; ***
= 1
per
cent
(tw
o-ta
iled
test
). T
hest
atis
tical
sig
nifi
canc
e le
vels
of
the
diff
eren
ces
betw
een
impa
cts
are
not n
oted
in th
e fi
gure
. Non
e of
the
diff
eren
ces
betw
een
impa
cts
for
the
child
ren
aged
3-5
and
the
child
ren
aged
6-9
wer
e st
atis
tical
ly s
igni
fica
nt.
In M
FIP,
beh
avio
r pr
oble
ms
wer
e m
easu
red
usin
g pa
rent
s' r
espo
nses
to a
12-
item
ext
erna
lizin
g su
bsca
le o
f th
e B
ehav
iora
l Pro
blem
s In
dex
that
ass
esse
s ag
gres
sive
beha
vior
s su
ch a
s bu
llyin
g an
d ch
eatin
g. R
espo
nses
ran
ge f
rom
0 (
"not
true
") to
2 (
"ver
y tr
ue")
. The
res
pons
es to
the
12 q
uest
ions
wer
e su
mm
ed to
com
pute
a s
ingl
e sc
ore
for
each
child
.In
SSP
, beh
avio
r pr
oble
ms
wer
e m
easu
red
usin
g a
four
-ite
m e
xter
naliz
ing
subs
cale
that
ask
ed p
aren
ts to
ass
ess
thei
r ch
ild's
act
ing
out a
nd a
ggre
ssiv
e be
havi
ors
on at
hree
poin
t sca
le r
angi
ng f
rom
1 (
"nev
er")
to 3
("o
ften
"). T
he r
espo
nses
wer
e av
erag
ed a
cros
s th
e fo
ur it
ems
to c
ompu
te a
sin
gle
scor
e fo
r ea
ch c
hild
.In
New
Hop
e, b
ehav
ior
prob
lem
s w
ere
mea
sure
d us
ing
a si
x-ite
m e
xter
naliz
ing
subs
cale
of
the
Prob
lem
Beh
avio
r Sc
ale
from
the
Soci
al S
kills
Rat
ing
Syst
em th
at a
sked
pare
nts
and
teac
hers
abo
ut th
e ch
ild's
agg
ress
ive
beha
vior
and
how
oft
en th
e ch
ild n
eede
d to
be
disc
iplin
ed f
or m
isbe
havi
or o
n a
five
-poi
nt s
cale
ran
ging
fro
m1
("ne
ver"
) to
5 (
"all
the
time"
). T
he r
espo
nses
wer
e av
erag
ed a
cros
s th
e si
x ite
ms
to c
ompu
te a
sin
gle
scor
e fo
r ea
ch c
hild
.
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
How
Wel
fare
and
Wor
k Po
licie
s A
ffec
t Chi
ldre
n
Figu
re 2
.8T
he E
ffec
ts o
n Sc
hool
Ach
ieve
men
t Did
Not
Dif
fer
for
Boy
s an
d G
irls
Con
sist
ently
Acr
oss
Ear
ning
s Su
pple
men
t Pro
gram
s
Pare
nts'
rep
orts
12
Tea
cher
s' r
epor
ts o
r ch
ildre
n's
test
sco
res
0.12
0.12
0.02
Boy
sG
irls
Boy
sG
irls
Enh
ance
d ea
rnin
gs d
isre
gard
(MFI
P In
cent
ives
Onl
y)C
ash
supp
lem
ent f
orfu
ll-tim
e w
ork
(SSP
)
0.04
Boy
sG
irls
Cas
h su
pple
men
t and
oth
er s
ubsi
dies
Enh
ance
d ea
rnin
gs d
isre
gard
with
for
full-
time
wor
k (N
ew H
ope)
wor
k m
anda
te (
Full
MFI
P)
NO
TE
S: I
n ea
ch s
tudy
, chi
ldre
n w
ere
sele
cted
for
incl
usio
n in
the
sam
ple
on th
e ba
sis
of th
eir
age
at r
ando
m a
ssig
nmen
t or
thei
r ag
e at
follo
w-u
p.T
he M
FIP
sam
ple
incl
udes
chi
ldre
n of
par
ents
in th
e M
FIP
eval
uatio
n ag
ed 5
-12
at th
e tim
e of
the
thre
e-ye
ar f
ollo
w-u
p su
rvey
(ag
ed a
ppro
xim
atel
y 2-
9 at
the
time
ofra
ndom
ass
ignm
ent)
who
se p
aren
ts w
ere
long
-ter
m r
ecip
ient
s in
urb
an c
ount
ies
and
unde
rwen
t ran
dom
ass
ignm
ent b
etw
een
Apr
il 1,
199
4, a
nd O
ctob
er 3
1, 1
994
(sam
ple
size
for
Full
MFI
P =
587
; sam
ple
size
for
MFI
P In
cent
ives
Onl
y =
573
).T
he S
SP s
ampl
e in
clud
es c
hild
ren
(of
pare
nts
in th
e SS
P ev
alua
tion)
age
d 6-
11 a
t the
tim
e of
the
thre
e-ye
ar s
urve
y (a
ged
appr
oxim
atel
y 3-
8 at
ran
dom
assi
gnm
ent)
who
wer
e liv
ing
in th
e ho
me
at th
e tim
e of
ran
dom
ass
ignm
ent a
nd a
t the
tim
e of
the
thre
e-ye
ar f
ollo
w-u
p su
rvey
(sa
mpl
e si
ze =
2,1
58).
The
New
Hop
e sa
mpl
e in
clud
es c
hild
ren
of th
e si
ngle
par
ents
in th
e N
ew H
ope
eval
uatio
n w
ho w
ere
aged
1-1
0 at
ran
dom
ass
ignm
ent a
nd w
hose
par
ents
part
icip
ated
in th
etw
o-ye
ar f
ollo
w-u
p su
rvey
(sa
mpl
e si
ze =
832
).T
he s
tatis
tical
sig
nifi
canc
e le
vels
of
the
impa
cts
are
indi
cate
d as
: * =
10
perc
ent;
** =
5 p
erce
nt; *
** =
1 p
erce
nt (
two-
taile
d te
st).
The
stat
istic
al s
igni
fica
nce
leve
ls o
f th
edi
ffer
ence
s be
twee
n im
pact
s ar
e no
t not
ed in
the
figu
re. T
he o
nly
diff
eren
ce b
etw
een
impa
cts
that
was
sta
tistic
ally
sig
nifi
cant
was
the
one
for
achi
evem
ent i
n M
FIP
Ince
ntiv
es O
nly.
In M
FIP,
ach
ieve
men
t was
ass
esse
d us
ing
a si
ngle
-ite
m m
easu
re th
at a
sked
par
ents
to r
ate
thei
r ch
ild's
ove
rall
perf
orm
ance
in s
choo
l on
a sc
ale
rang
ing
from
1 ("
not w
ell
at a
ll")
to 5
("v
ery
wel
l").
In S
SP, a
chie
vem
ent w
as m
easu
red
usin
g a
26-
to 3
4-ite
m m
ath
skill
s te
st a
nd e
xpre
ssed
in te
rms
of th
e pr
opor
tion
of it
ems
answ
ered
cor
rect
ly. P
aren
ts' a
sses
smen
ts o
fac
hiev
emen
t wer
e m
easu
red
usin
g th
eir
ratin
gs o
f th
eir
child
's f
unct
ioni
ng in
thre
e ac
adem
ic s
ubje
cts
on a
fiv
e-po
int s
cale
ran
ging
fro
m I
("n
ot v
ery
wel
l")
to5
("ve
ry w
ell"
). T
hera
tings
wer
e av
erag
ed a
cros
s th
e th
ree
acad
emic
sub
ject
s to
com
pute
a s
ingl
e sc
ore
for
each
chi
ld.
In N
ew H
ope,
teac
hers
' rep
orts
of
achi
evem
ent w
ere
mea
sure
d us
ing
the
10-i
tem
Aca
dem
ic S
ubsc
ale
from
the
Soci
al S
kills
Rat
ing
Syst
em, w
hich
ask
ed te
ache
rs to
rat
eth
e ch
ild's
ski
lls r
elat
ive
to th
ose
of o
ther
chi
ldre
n in
are
as s
uch
as m
ath,
rea
ding
, and
ora
l com
mun
icat
ion
on a
fiv
e-po
int s
cale
ran
ging
fro
m 1
("bo
ttom
10
perc
ent"
) to
5 (
"top
10
perc
ent"
). T
he r
espo
nses
wer
e av
erag
ed a
cros
s th
e 10
item
s to
com
pute
a s
ingl
e sc
ore
for
each
chi
ld. P
aren
ts' a
sses
smen
ts o
f ac
hiev
emen
t wer
e m
easu
red
usin
g a
sing
le-i
tem
mea
sure
that
ask
ed p
aren
ts to
rat
e th
eir
child
's s
choo
l per
form
ance
, bas
ed o
n pa
st r
epor
t car
ds o
r ot
her
sour
ces,
on
a fi
ve-p
oint
sca
le r
angi
ng f
rom
1("
not a
t all
wel
l")
to 5
("v
ery
wel
l").
Full MFIP were larger for boys than for girls (although the differences between the subgroup im-pacts were not statistically significant in any of the latter three programs).
As for children's behavior outcomes, two of the earnings supplement programs had amore favorable impact on girls' behavior than on boys', while one program had no effect for ei-ther boys or girls and one had favorable effects for boys but unfavorable effects for girls (seeFigure 2.9). In the two MFIP programs there were reductions in behavior problems for girls butnot for boys (although only in MFIP Incentives Only was the difference between impacts statisti-cally significant). SSP had no effect on behavior problems for boys or girls. New Hope had morefavorable impacts on behavior problems for boys than for girls (see Figure 2.9). In fact, the pro-gram's impacts on girls' behavior problems were unfavorable: Teachers rated girls in the pro-gram group as having more behavior problems than girls in the control group. Finally, the effectson positive behavior are also inconsistent across programs. For instance, whereas MFIP Incen-tives Only had positive effects for girls but not for boys, New Hope had positive effects for boysbut not for girls (not shown in figures).
In sum, some of the programs with earnings supplements had effects primarily on girls,while others had effects primarily on boys. With one exception, all these effects were either fa-vorable or neutral. There is no clear pattern across programs suggesting that one gender is fa-vored over the other.
IX. Summary and Discussion of the Effects of Programswith Earnings Supplements
These findings suggest that programs with earnings supplements lead to small to modestimprovements in child outcomes, particularly for children in long-term recipient families. Thefeature that these programs share is the offer of a generous earnings supplement; their differencesin case management, employment services, in-kind benefits, and whether they provide earningssupplements within or outside the welfare system do not appear to result in differences betweenthese programs in their impacts on children. Moreover, the results appear to hold across a diverseset of sites, samples, and macroeconomic conditions.
Can these findings be generalized to other programs that include earnings supplements?The earnings supplements offered by the programs examined in this chapter were generous, butnot more so than those offered by a number of programs currently in effect.
As discussed in greater detail in the companion monograph13 focusing on the impacts ofwelfare and employment policies on parental employment and income, a typical program groupmember in the MFIP study who worked 20 hours per week at $6 an hour took home about $250more in monthly income than she would have under the old welfare rules (which applied to thecontrol group). If she worked 40 hours per week, however, her monthly income would have beenonly about $150 higher than it would have been under the old rules. In SSP, in contrast, a typicalprogram group member working 20 hours per week at $6 an hour who received the earnings sup-plement did not receive any more income than a typical control group member who worked the
I3Bloom and Michalopoulos, 2001.
-39-5 9
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
-0.0
5
-0.1
0
-0.1
5
-0.2
0
-0.2
5
-0.3
0
-0.3
5
-0.4
0
-0.4
5
-0.5
0
How
Wel
fare
and
Wor
k Po
licie
s A
ffec
t Chi
ldre
n
Figu
re 2
.9T
he E
ffec
ts o
n B
ehav
ior
Prob
lem
s D
id N
ot D
iffe
r fo
r B
oys
and
Gir
lsC
onsi
sten
tly A
cros
s E
arni
ngs
Supp
lem
ent P
rogr
ams
Pare
nts'
rep
orts
II
Tea
cher
s' r
epor
ts
Boy
sG
irls
Boy
sG
irls
Enh
ance
d ea
rnin
gs d
isre
gard
Cas
h su
pple
men
t for
ful
l-tim
e(M
FIP
Ince
ntiv
es O
nly)
wor
k (S
SP)
-0.1
0
-0.2
2 *
Boy
sG
irls
Cas
h su
pple
men
t and
oth
er s
ubsi
dies
Enh
ance
d ea
rnin
gs d
isre
gard
with
for
full-
time
wor
k (N
ew H
ope)
wor
k m
anda
te (
Full
MFI
P)
NO
TE
S: I
n ea
ch s
tudy
, chi
ldre
n w
ere
sele
cted
for
incl
usio
n in
the
sam
ple
on th
e ba
sis
of th
eir
age
at r
ando
m a
ssig
nmen
t or
thei
r ag
e at
fol
low
-up.
The
MFI
P sa
mpl
e in
clud
es c
hild
ren
of p
aren
ts in
the
MFI
P ev
alua
tion
aged
5-1
2 at
the
time
of th
e th
ree-
year
fol
low
-up
surv
ey (
aged
app
roxi
mat
ely
2-9
at th
e tim
e of
rand
om a
ssig
nmen
t) w
hose
par
ents
wer
e lo
ng-t
erm
rec
ipie
nts
in u
rban
cou
ntie
s an
d un
derw
ent r
ando
m a
ssig
nmen
t bet
wee
n A
pril
1, 1
994,
and
Oct
ober
31,
1994
(sa
mpl
e si
ze f
orFu
ll M
FIP
= 5
87; s
ampl
e si
ze f
or M
FIP
Ince
ntiv
es O
nly
= 5
73).
The
SSP
sam
ple
incl
udes
chi
ldre
n of
sin
gle
pare
nts
in th
e SS
P ev
alua
tion
aged
6-1
1 at
the
time
of th
e th
ree-
year
fol
low
-up s
urve
y (a
ged
appr
oxim
atel
y 3-
8 at
ran
dom
assi
gnm
ent)
who
wer
e liv
ing
in th
e ho
me
at th
e tim
e of
ran
dom
ass
ignm
ent a
nd a
t the
tim
e of
the
thre
e-ye
ar f
ollo
w-u
psu
rvey
(sa
mpl
e si
ze =
2,1
58).
The
New
Hop
e sa
mpl
e in
clud
es c
hild
ren
of th
e si
ngle
par
ents
in th
e N
ew H
ope
eval
uatio
n w
ho w
ere
aged
1-1
0 at
ran
dom
ass
ignm
ent a
nd w
hose
pare
nts
part
icip
ated
inth
e tw
o-ye
ar f
ollo
w-u
p su
rvey
(sa
mpl
e si
ze =
832
).
The
sta
tistic
al s
igni
fica
nce
leve
ls o
f th
e im
pact
s ar
e in
dica
ted
as: *
= 1
0 pe
rcen
t; **
= 5
per
cent
; ***
= 1
per
cent
(tw
o-ta
iled
test
). T
hest
atis
tical
sig
nifi
canc
e le
vels
of
the
diff
eren
ces
betw
een
impa
cts
are
not n
oted
in th
e fi
gure
. The
onl
y di
ffer
ence
s be
twee
n im
pact
s th
at w
ere
stat
istic
ally
sig
nifi
cant
wer
e th
at in
par
ent-
repo
rted
beh
avio
r pr
oble
ms
for
MFI
P In
cent
ives
Onl
y an
d th
at in
teac
her-
repo
rted
beh
avio
r pr
oble
ms
for
New
Hop
e.In
MFI
P, b
ehav
ior
prob
lem
s w
ere
mea
sure
d us
ing
pare
nts'
res
pons
es to
a 1
2-ite
m e
xter
naliz
ing
subs
cale
of
the
Beh
avio
ral P
robl
ems
Inde
x th
atas
sess
es a
ggre
ssiv
ebe
havi
ors
such
as
bully
ing
and
chea
ting.
Res
pons
es r
ange
fro
m 0
("n
ot tr
ue "
) to
2 (
"ver
y tr
ue")
. The
res
pons
es to
the
12 q
uest
ions
wer
e su
mm
ed to
com
pute
a s
ingl
e sc
ore
for
each
child
.
In S
SP, b
ehav
ior
prob
lem
s w
ere
mea
sure
d us
ing
a fo
ur-i
tem
ext
erna
lizin
g su
bsca
le th
at a
sked
par
ents
to a
sses
s th
eir
child
's a
ctin
g ou
t and
agg
ress
ive
beha
vior
s on
a th
ree-
poin
t sca
le r
angi
ng f
rom
1 (
"nev
er")
to 3
("o
ften
"). T
he r
espo
nses
wer
e av
erag
ed a
cros
s th
e fo
ur it
ems
to c
ompu
te a
sin
gle s
core
for
eac
h ch
ild.
In N
ew H
ope,
beh
avio
r pr
oble
ms
wer
e m
easu
red
usin
g a
six-
item
ext
erna
lizin
g su
bsca
le o
f th
e Pr
oble
m B
ehav
ior
Scal
e fr
om th
e So
cial
Ski
lls R
atin
g Sy
stem
that
ask
edpa
rent
s an
d te
ache
rs a
bout
the
child
's a
ggre
ssiv
e be
havi
or a
nd h
ow o
ften
the
child
nee
ded
to b
e di
scip
lined
for
mis
beha
vior
on
a fi
ve-p
oint
sca
le r
angi
ng f
rom
I ("
neve
r")
to 5
("a
llth
e tim
e").
The
res
pons
es w
ere
aver
aged
acr
oss
the
six
item
s to
com
pute
a s
ingl
e sc
ore
for
each
chi
ld.
6263
same amount; if the program group member worked 40 hours per week at the same wage, how-ever, she received nearly $400 more per month than the control group member.
How do these supplements compare with those provided by programs now in effect? Thefederal Earned Income Credit (EIC) currently provides nearly $4,000 per year to a parent withtwo children who works full time at a minimum-wage job. In addition, most states have imple-mented an "enhanced earnings disregard" as part of their welfare reform strategy, as did theMFIP programs. In a number of states, the enhanced earnings disregards are as generous as thesupplements examined here, or more so. A welfare recipient in Connecticut, for instance, cannow continue receiving all her welfare and Food Stamp benefits as long as she earns less than thefederal poverty threshold. Relative to how she would have fared under the AFDC system, thisdisregard provides her with about $600 more per month in income if she works full time at aminimum-wage job. And California now allows welfare recipients who work to keep the first$225 of their monthly earnings without having their welfare benefits reduced; beyond that point,each additional dollar of earnings reduces their benefits by only half a dollar (rather than reduc-ing benefits by about a dollar for every dollar of earnings as under AFDC). As a result, a workingwelfare recipient in California can receive as much of an income boost as a program group mem-ber who received the maximum benefits in these studies. The situation is similar in other high-grant states that have expanded their earnings disregards.
At the same time, not all enhanced disregards are as generous as the supplements pro-vided by the programs analyzed in this chapter. In some states, the disregard is very low, some-times as low as 20 percent of a recipient's earnings (in Nebraska, for example). Also, in stateswith very low benefit levels, even an enhanced earnings disregard translates into little increase infamily income. For example, in Connecticut, where the maximum benefit for a family of three isover $500, 100 percent of income from earnings is disregarded in calculating the grant level. InNorth Carolina, the earnings disregard in the first three months of employment is the same as inConnecticut, but the corresponding welfare grant is half as large. Therefore, the boost in incomefrom remaining on welfare while working is much smaller in North Carolina than in Connecticut.
The findings in this monograph suggest that make-work-pay strategies benefit children.But it is unclear whether programs that provide less generous earnings supplements will have thesame benefits for children as the earnings supplement programs examined here.
-42-
Chapter 3
Effects on Children of Programs with MandatoryEmployment Services
Mandatory employment services are intended to increase employment by requiring par-ents to participate in employment, employment-related activities, or education as a condition ofreceiving welfare benefits. On the one hand, mandatory employment services may increase pa-rental stress particularly when parents feel coerced to participate and thereby affect chil-dren adversely. A large body of data indicates that maternal employment is associated with fewerpositive and more negative outcomes for children when mothers believe that it is inconsistentwith their roles as parents.' Moreover, parents who fail to participate in mandatory programs maybe faced with a partial or complete loss of welfare benefits, which may increase stress by reduc-ing family income. On the other hand, because such programs may increase employment and re-duce welfare receipt, they may improve parents' sense of competence and well-being, whichcould result in positive outcomes for children. In Chapter 2, we found that adding a participationmandate did not diminish the positive effects on children of an earnings supplement program. Inthis chapter, we focus on the effects of the six out of the 11 programs in this monograph that pro-vided mandatory employment services without earnings supplements; all six programs were stud-ied in the NEWWS evaluation.2
Each of the six programs took one of two basic approaches to providing mandatory em-ployment services: a job-search-first approach or an education-first approach. In the job-search-first approach, welfare recipients were required to begin by looking for work, either on their ownor through group activities that teach job-seeking skills (such as job clubs). In the education-firstapproach, most participants were assigned initially to classroom-based education or training ser-vices. For participants who had not completed high school, these classes took the form of basiceducation programs offering remedial English or math instruction or preparation for the GeneralEducational Development (GED) exam. For high school graduates, these classes took the form ofvocational educational programs. Notably, the education provided in these programs was gener-ally not at an advanced or college level but was instead basic educational instruction that wasgenerally available in the community. Therefore, the results cannot tell us how children are af-fected by programs that provide parents with higher-level education.
I. Effects on Parents' Economic Outcomes
The programs with mandatory employment services generally increasedemployment.
The impacts on employment of the programs that provided mandatory employment services(without earnings supplements) are comparable to those of the earnings supplement programs dis-
'Alvarez, 1985; Farel, 1980.2This chapter is based on a reanalysis of the data from the NEWWS Child Outcomes Study (McGroder et al., 2000).
65-43-
cussed in Chapter 2.3 For families with young children (the sample analyzed here), five of the sixprograms increased employment.4 Because the education-first programs initially placed parents inadult basic education, their impacts on employment in the first year were generally smaller thanthose of job-search-first programs. This difference seems to, have disappeared by the third year, al-though it nevertheless resulted in larger average increases in earnings for the job-search-first pro-grams compared with the education-first programs over the evaluation period as a whole.5
In addition to increasing employment, mandatory employment servicesare intended to reduce welfare receipt, and the programs with mandatesexamined here did so. However, the programs generally left income un-changed.
The effects of the programs with mandatory employment services on welfare receipt gen-erally mirror their effects on employment: Those that had the largest effects on employment re-duced welfare receipt the most.6 Four of the six programs examined here decreased welfare re-ceipt among families with young children.' Because parents are in essence trading their welfarebenefits for earnings, however, families in these programs were left with no more income on av-erage than families in the control group, who were more likely to be receiving welfare.
Unlike the programs with earnings supplements, the programs that pro-vided only mandatory employment services cost little and sometimessaved the government money.
The average net cost over two years of the six programs with mandatory employment ser-vices ranged from a savings of $1,678 per family to a cost of $2,968 per family, lower than theaverage net cost of the programs with earnings supplements. Because of the welfare savings gen-erated by these programs, mandatory employment services by themselves increase family earn-ings but not family income, and they can sometimes save the government money.
II. Effects on Children
The programs with mandatory employment services generally had no ef-fect on young children's school achievement.
Figure 3.1 shows the effects on young children's school achievement of each of the edu-cation-first and job-search-first programs in the NEWWS evaluation (see left panel) alongsidethe effects of the programs with earnings supplements discussed in Chapter 2 (see right panel).Because all the children in the NEWWS Child Outcomes Study8 (also the sample of children ex-amined here) were 3-5 years old when their parents underwent random assignment, for purposesof comparison the results for the earnings supplement programs shown in Figure 3.1 are limited
3Bloom and Michalopoulos, 2001.4Hamilton, 2000.5Bloom and Michalopoulos, 2001.6Bloom and Michalopoulos, 2001.'Hamilton, 2000.8McGroder et al., 2000. 66
-44-
How
Wel
fare
and
Wor
k Po
licie
s A
ffec
t Chi
ldre
n
Figu
re 3
.1Fo
r C
hild
ren
Age
d 3-
5 at
Ran
dom
Ass
ignm
ent,
Prog
ram
s w
ith M
anda
tory
Em
ploy
men
t Ser
vice
sH
ad F
ew E
ffec
ts o
n Sc
hool
Ach
ieve
men
t
Pare
nts'
rep
orts
Tea
cher
s' r
epor
ts o
r ch
ildre
n's
test
sco
res
0.40
0.35
0.30
0.25
Prog
ram
s w
ith m
anda
tory
em
ploy
men
t ser
vice
s on
lyPr
ogra
ms
with
ear
ning
s su
pple
men
ts
0.31
**
0.22
**
0.19
***
0.20
**0.
150.
150.
100.
100.
07
I0.1
5
0.05
0.03
0.03
0.03
0.01
0.00
-0.0
5-0
.04
Atla
nta
Gra
nd R
apid
sR
iver
side
Atla
nta
Gra
nd R
apid
sR
iver
side
MFI
PSS
PN
ew H
ope
Full
MFI
PN
EW
WS
NE
WW
SN
EW
WS
NE
WW
SN
EW
WS
NE
WW
SIn
cent
ives
Onl
y
Edu
catio
n-fi
rst p
rogr
ams
Job-
sear
ch-f
irst
pro
gram
s
NO
TE
S: I
n ea
ch s
tudy
, chi
ldre
n w
ere
sele
cted
for
incl
usio
n in
the
sam
ple
on th
e ba
sis
of th
eir
age
at r
ando
m a
ssig
nmen
t or
thei
r ag
e at
follo
w-u
p.T
he N
EW
WS
sam
ple
incl
udes
chi
ldre
n of
sin
gle
mot
hers
in th
e N
EW
WS
eva
luat
ion
aged
3-5
at t
he b
egin
ning
of t
he s
tudy
who
se p
aren
tsw
ere
rand
omly
sel
ecte
d to
par
ticip
ate
in th
e tw
o-ye
ar fo
llow
-up
surv
ey (
sam
ple
size
s fo
r ed
ucat
ion-
first
pro
gram
s: A
tlant
a =
1,0
26, G
rand
Rap
ids
= 4
21, R
iver
side
= 5
78; s
ampl
e si
zes
for
job-
sear
ch-f
irst p
rogr
ams:
Atla
nta
= 9
02, G
rand
Rap
ids
= 4
41, R
iver
side
= 6
94).
The
MF
IP s
ampl
e in
clud
es c
hild
ren
of p
aren
ts in
the
MF
IP e
valu
atio
n ag
ed 6
-8 a
t the
tim
e of
the
thre
e-ye
ar fo
llow
-up
surv
ey (
aged
app
roxi
mat
ely
3-5
atth
e tim
e of
ran
dom
ass
ignm
ent)
who
se p
aren
ts w
ere
long
-ter
m r
ecip
ient
s in
urb
an c
ount
ies
and
unde
rwen
t ran
dom
ass
ignm
ent b
etw
een
Apr
il 1,
199
4, a
nd O
ctob
er 3
1, 1
994
(sam
ple
size
for
Ful
l MF
IP=
286
; sam
ple
size
for
MF
IP In
cent
ives
Onl
y =
289
).T
he S
SP
sam
ple
incl
udes
chi
ldre
n of
sin
gle
pare
nts
in th
e S
SP
eva
luat
ion
aged
6-8
at t
he ti
me
of th
e th
ree-
year
follo
w-u
p su
rvey
(ag
ed a
ppro
xim
atel
y 3-
5at
ran
dom
ass
ignm
ent)
who
wer
e liv
ing
in th
e ho
me
at r
ando
m a
ssig
nmen
t and
at t
he ti
me
of th
e th
ree-
year
follo
w-u
p su
rvey
(sa
mpl
e si
ze =
1,3
18).
The
New
Hop
e sa
mpl
e in
clud
es c
hild
ren
of th
e si
ngle
par
ents
in th
e N
ew H
ope
eval
uatio
n w
ho w
ere
aged
3-5
at r
ando
m a
ssig
nmen
t and
who
se p
aren
tspa
rtic
ipat
ed in
the
two-
year
follo
w-u
p su
rvey
(sa
mpl
esi
ze =
265
). Sta
tistic
al s
igni
fican
ce le
vels
are
indi
cate
d as
: * =
10
perc
ent;
** =
5 p
erce
nt; *
** =
1 p
erce
nt (
two-
taile
d te
st).
In N
EW
WS
, ach
ieve
men
t was
mea
sure
d us
ing
child
ren'
s st
anda
rd s
core
s on
the
Bra
cken
Sch
ool R
eadi
ness
Com
posi
te te
st, w
hich
ass
esse
s kn
owle
dge
of c
olor
s, le
tters
, num
bers
/cou
ntin
g,co
mpa
rison
s, a
ndsh
apes
In M
FIP
, ach
ieve
men
t was
ass
esse
d us
ing
a si
ngle
-item
mea
sure
that
ask
ed p
aren
ts to
rat
e th
eir
child
's o
vera
ll pe
rfor
man
ce in
sch
ool o
n a
scal
e ra
ngin
g fr
om I
("no
t wel
l at a
ll")
to 5
("v
ery
wel
l").
In S
SP
, ach
ieve
men
t was
mea
sure
d us
ing
a 26
- to
34-
item
mat
h sk
ills
test
and
exp
ress
ed in
term
s of
the
prop
ortio
n of
item
s an
swer
ed c
orre
ctly
. Par
ents
' ass
essm
ents
of a
chie
vem
ent w
ere
mea
sure
d us
ing
thei
r ra
tings
of t
heir
child
's fu
nctio
ning
in th
ree
acad
emic
sub
ject
s on
a fi
ve-p
oint
sca
le r
angi
ng fr
om I
("no
t ver
y w
ell")
to .5
("v
ery
wel
l"). T
hera
tings
wer
e av
erag
ed a
cros
s th
e th
ree
acad
emic
sub
ject
s to
com
pute
asi
ngle
sco
re fo
r ea
ch c
hild
.
In N
ew H
ope,
teac
hers
' rep
orts
of a
chie
vem
ent w
ere
mea
sure
d us
ing
the
10-it
em A
cade
mic
Sub
scal
e fr
om th
e S
ocia
l Ski
lls R
atin
g S
yste
m, w
hich
ask
ed te
ache
rsto
rat
e th
e ch
ild's
ski
lls r
elat
ive
to th
ose
ofot
her
child
ren
in a
reas
suc
h as
mat
h, r
eadi
ng, a
nd o
ral c
omm
unic
atio
n on
a fi
ve-p
oint
sca
le r
angi
ng fr
om 1
("b
otto
m 1
0 pe
rcen
t")
to 5
("t
op 1
0 per
cent
"). T
he r
espo
nses
wer
e av
erag
ed a
cros
s th
e 10
item
s to
com
pute
asi
ngle
sco
re fo
r ea
ch c
hild
. Par
ents
' ass
essm
ents
of a
chie
vem
ent w
ere
mea
sure
d us
ing
a si
ngle
-item
mea
sure
that
ask
ed p
aren
ts to
rat
e th
eir
child
's s
choo
l per
form
ance
, bas
edon
pas
t rep
ort c
ards
or
othe
r so
urce
s, o
n a
five-
poin
t sca
le r
angi
ng fr
om I
("no
t at a
ll w
ell")
to 5
("v
ery
wel
l").
67
to the children who were 3-5 years old at the time of their parents' random assignment. Unlikethe programs with earnings supplements, which had generally positive impacts on achievement,
the programs that provided only mandatory employment services produced few effects. Only oneof the six programs affected test scores at all: Children of parents in Atlanta's jobs-search-firstprogram had higher test scores, on average, than their control group counterparts.
The effects on children's school achievement of programs with manda-tory employment services did not differ for job-search-first 'programscompared with education-first programs.
How are children affected by programs that use a job-search-first as opposed to an educa-tion-first strategy? To answer this question, the NEWWS evaluation was designed to allow side-by-side comparison of the two approaches in each of the three sites. One might expect that mov-ing parents into employment would have a different effect on children's achievement than mov-ing parents into education programs. Although in both cases children are faced with separationfrom their parents, parents who are engaged in school programs may serve as role models con-veying the importance of school to their children and may become better teachers to their chil-
dren. In that case, then one would predict education-first programs to have more positive effectsthan job-search-first programs on children's school achievement. Indeed, studies have found thatparents with more education use a more positive teaching style with their children.9 However,studies that attempted to control for other characteristics associated with differences in mothers'education level (such as income) found that a higher level of education had mixed effects onchildren's achi evement. 1 °
As Figure 3.1 makes evident, neither the education-first nor the job-search-first approachaffected children's school achievement consistently across the sites in NEWWS. Both ap-proaches produced small and generally not statistically significant effects on test scores. The ef-
fects that were observed appear to be specific to particular sites rather than associated with a par-ticular program approach. In other words, there was greater similarity between the effects of pro-grams in the same site than between the effects of programs that used the same approach.
The programs with mandatory employment services generally had mixedeffects on children's behavior; the effects on children's health were neu-tral or negative.
As Figure 3.2 illustrates, the pattern of effects on children's behavior problems is notconsistent across the six programs with mandatory employment services. Both Atlanta programsreduced children's behavior problems." In Grand Rapids, however, the education-first and job-search-first programs had effects of the same magnitude but in the opposite direction, actuallyincreasing behavior problems. These effects seem to be linked to specific sites rather than to spe-
9Laosa, 1983.I°Kaestrier and Corman, 1995; Rosenweig and Wolpin, 1994."Analyses conducted as part of the NEWWS Child Outcomes Study (McGroder et al., 2000) found that the re-
duction in behavior problems in the Atlanta education-first program was in the same direction but not statisticallysignificant. Minor differences between the covariates included in the statistical models used there and here may ac-count for this slight discrepancy between findings.
69-46-
How
Wel
fare
and
Wor
k Po
licie
s A
ffec
t Chi
ldre
n
Figu
re 3
.2Fo
r C
hild
ren
Age
d 3-
5 at
Ran
dom
Ass
ignm
ent,
Prog
ram
s w
ith M
anda
tory
Em
ploy
men
t Ser
vice
sD
id N
ot C
onsi
sten
tly R
educ
e or
Inc
reas
e B
ehav
ior
Prob
lem
s
OPa
rent
s' r
epor
ts M
Tea
cher
s' r
epor
ts
Prog
ram
s w
ith m
anda
tory
em
ploy
men
t ser
vice
s on
ly
0.25
0.20
0.15
0.10
0.15
0.17
0.05
0.05
0.02
0.00
-0.0
5
-0.1
0-0
.10
*-0
.15
-0.1
5 **
*-0
.20
Atla
nta
Gra
nd R
apid
sR
iver
side
Atla
nta
Gra
nd R
apid
sR
iver
side
NE
WW
SN
EW
WS
NE
WW
SN
EW
WS
NE
WW
SN
EW
WS
Edu
catio
n-fi
rst p
rogr
ams
Job-
sear
ch-f
irst
pro
gram
s
Prog
ram
s w
ith e
arni
ngs
supp
lem
ents
0.01 MFI
PIn
cent
ives
Onl
y
0.01
-0.1
9SS
PN
ew H
ope
Full
MFI
P
NO
TE
S: I
n ea
ch s
tudy
, chi
ldre
n w
ere
sele
cted
for
incl
usio
n in
the
sam
ple
on th
e ba
sis
of th
eir
age
at r
ando
m a
ssig
nmen
t or
thei
r ag
e at
follo
w-u
p.T
he N
EW
WS
sam
ple
incl
udes
chi
ldre
n of
sin
gle
mot
hers
in th
e N
EW
WS
eva
luat
ion
aged
3-5
at t
he b
egin
ning
of t
he s
tudy
who
se p
aren
ts w
ere
rand
omly
sel
ecte
d to
part
icip
ate
in th
e tw
o-ye
ar fo
llow
-up
surv
ey (
sam
ple
size
s fo
r ed
ucat
ion-
first
pro
gram
s: A
tlant
a 1,
026,
Gra
nd R
apid
s =
421
, Riv
ersi
de =
578
; sam
ple
size
s fo
r jo
b-se
arch
-firs
t pro
gram
s: A
tlant
a =
902
, Gra
nd R
apid
s = 4
41, R
iver
side
= 6
94).
The
MF
IP s
ampl
e in
clud
es c
hild
ren
of p
aren
ts in
the
MF
IP e
valu
atio
n ag
ed 6
-8 a
t the
tim
e of
the
thre
e-ye
ar fo
llow
-up
surv
ey (
aged
app
roxi
mat
ely
3-5
atth
e tim
e of
ran
dom
ass
ignm
ent)
who
se p
aren
ts w
ere
long
-ter
m r
ecip
ient
s in
urb
an c
ount
ies
and
unde
rwen
t ran
dom
ass
ignm
ent b
etw
een
Apr
il 1,
199
4, a
nd O
ctob
er 3
1, 1
994
(sam
ple
size
for
Ful
l MF
IP =
286
; sam
ple
size
for
MF
IP In
cent
ives
Onl
y =
289
).T
he S
SP
sam
ple
incl
udes
chi
ldre
n of
sin
gle
pare
nts
in th
e S
SP
eva
luat
ion
aged
6-8
at t
he ti
me
of th
e th
ree-
year
follo
w-u
p su
rvey
(ag
ed a
ppro
xim
atel
y 3-
5 at r
ando
m a
ssig
nmen
t) w
ho w
ere
livin
g in
the
hom
eat
ran
dom
ass
ignm
ent a
nd a
t the
tim
e of
the
thre
e-ye
ar fo
llow
-up
surv
ey (
sam
ple
size
= 1
,318
).
The
New
Hop
e sa
mpl
e in
clud
es c
hild
ren
of th
e si
ngle
par
ents
in th
e N
ew H
ope
eval
uatio
n w
ho w
ere
aged
3-5
at r
ando
m a
ssig
nmen
t and
who
se p
aren
tspa
rtic
ipat
ed in
the
two-
year
follo
w-u
p su
rvey
(sa
mpl
esi
ze =
265
). Sta
tistic
al s
igni
fican
ce le
vels
are
indi
cate
d as
: * =
10
perc
ent;
** =
5 p
erce
nt; *
** =
1 p
erce
nt (
two-
taile
d te
st).
In N
EW
WS
, beh
avio
r pr
oble
ms
wer
e m
easu
red
usin
g pa
rent
s' r
espo
nses
to a
12-
item
ext
erna
lizin
g su
bsca
le o
f the
Beh
avio
ral P
robl
ems
Inde
x, w
hich
ass
esse
s ag
gres
sive
beh
avio
rs s
uch
asbu
llyin
g an
dch
eatin
g. R
espo
nses
ran
ge fr
om 0
("n
ot tr
ue")
to 2
("v
ery
true
"). T
he r
espo
nses
to th
e 12
que
stio
ns w
ere
sum
med
to c
ompu
te a
sin
gle
scor
e fo
rea
ch c
hild
.In
MF
IP, b
ehav
ior
prob
lem
s w
ere
mea
sure
d us
ing
pare
nts'
res
pons
es to
a 1
2-ite
m e
xter
naliz
ing
subs
cale
of t
he B
ehav
iora
l Pro
blem
s In
dex
that
ass
esse
s ag
gres
sive
beh
avio
rs s
uch
asbu
llyin
g an
d ch
eatin
g.R
espo
nses
ran
ge fr
om 0
("n
ot tr
ue")
to 2
("v
ery
true
"). T
he r
espo
nses
to th
e 12
que
stio
ns w
ere
sum
med
to c
ompu
te a
sin
gle
scor
e fo
r ea
chch
ild.
In S
SP
, beh
avio
r pr
oble
ms
wer
e m
easu
red
usin
g a
four
-item
ext
erna
lizin
g su
bsca
le th
at a
sked
par
ents
to a
sses
s th
eir
child
's a
ctin
g ou
t and
agg
ress
ive
beha
vior
s on
ath
ree-
poin
t sca
le r
angi
ng fr
om 1
("ne
ver"
) to
3 (
"ofte
n").
The
res
pons
es w
ere
aver
aged
acr
oss
the
four
item
s to
com
pute
a s
ingl
e sc
ore
for
each
chi
ld.
-
In N
ew H
ope,
beh
avio
r pr
oble
ms
wer
e m
easu
red
usin
g a
six-
item
ext
erna
lizin
g su
bsca
le o
f the
Pro
blem
Beh
avio
r S
cale
from
the
Soc
ial S
kills
Rat
ing
Sys
tem
that
ask
edpa
rent
s an
d te
ache
rs a
bout
the
child
'sag
gres
sive
beh
avio
r an
d ho
w o
ften
the
child
nee
ded
to b
e di
scip
lined
for
mis
beha
vior
on
a fiv
e-po
int s
cale
ran
ging
from
1 (
"nev
er")
to 5
("a
ll th
etim
e").
The
res
pons
es w
ere
aver
aged
acr
oss
the
six
item
s to
com
pute
asi
ngle
sco
re fo
r ea
ch c
hild
. 7071
cific program approaches. Although the earnings supplement programs generated few effects onbehavior problems for children in this age group (see right panel of Figure 3.2), none of the earn-ings supplement programs had an unfavorable impact on this outcome.
Both Riverside programs had negative effects on children's health as measured by par-ents' ratings, but neither positive nor negative effects were found at the other two sites (see Fig-ure 3.3). In contrast, out of the three earnings supplement programs in which children's healthwas measured only SSP had positive effects on children's health.
The passage of the 1996 welfare legislation gave rise to considerable concern about howfamilies with long-term welfare receipt would fare. In Chapter 2, we reported that the positiveeffects of earnings supplement programs were more pronounced for children of long-term recipi-ents. Now we examine the effects of programs with mandatory employment services on children
in this same subgroup of families.
Analyses of children in long-term recipient families suggest that the mixed pattern ofthese programs' effects holds in this subgroup as well (see Appendix Table 6). In fact, for thelong-term recipient families the effects on children are generally stronger than in the full sampleof families analyzed here. However, unlike for the full sample, the direction of effects across out-comes varies by site for the long-term recipient families. For example, the Atlanta programs werefound to produce large improvements in children's achievement and reductions in children's be-havior problems for long-term welfare recipients. But in the same subgroup the Grand Rapidsprograms increased behavior problems, and the Riverside programs led to large negative effectson children's health.
In sum, programs that included mandatory employment services had feweffects on children, and the effects found were mixed.12
While few effects on children were found in the programs evaluated in NEWWS, those ofthe Atlanta programs (particularly the job-search-first program) tended to be positive, while thoseof the Grand Rapids and Riverside programs tended to be negative. This pattern suggests thatlocal economic conditions, the population served, or program implementation may influence howparents and their children respond to mandatory employment programs. For example, a closerexamination of the populations served by the six programs in NEWWS suggests that the popula-tion in Atlanta had much lower levels of reading and math skills than the populations in GrandRapids and Riverside.13 As a result, program implementers in the Atlanta job-search-first pro-gram focused more heavily on basic education than did program implementers in the other twosites' job-search-first programs. The Atlanta programs were also the most "customer oriented" ofthe six programs, with staff members emphasizing counseling and the benefits of the programmore than the threat of sanctions.14 The greater degree of disadvantage among the Atlanta par-
"Notably, this conclusion does not change if one examines the impacts on measures of the proportion of chil-dren performing above or below average on measures of child achievement, behavior, and health. There is no consis-tent pattern of effects across programs on the proportion of children performing at the top or bottom of the distribu-tion in these samples (McGroder et al., 2000).
"Hamilton, Brock, Farrell, Friedlander, and Harknett, 1997.I4Hamilton et al., 1997.
-48-72
How
Wel
fare
and
Wor
k Po
licie
s A
ffec
t Chi
ldre
n
Figu
re 3
.3Fo
r C
hild
ren
Age
d 3-
5 at
Ran
dom
Ass
ignm
ent,
Tw
o Pr
ogra
ms
with
Man
dato
ry E
mpl
oym
ent S
ervi
ces
Had
Neg
ativ
e E
ffec
ts o
n H
ealth
0.15
0.10
7 :
0.05
Z
Prog
ram
s
0.09
with
man
dato
ry e
mpl
oym
ent s
ervi
ces
only
Prog
ram
s w
ithea
rnin
gs
0.13
**
supp
lem
ents
0.00
a, -
0.05
E 4..
-0.1
00 a> 1.
1-0
.15
4.,
e.)
-0.2
0L
.
-0.2
5
-0.3
0
II
Rap
ids
NE
WW
S
1
-0.2
1
Riv
ersi
deIn
cent
ives
SSP
Full
Onl
y
-0.0
2
Atla
nta
Gra
ndN
EW
WS
-0.0
2
Atla
nta
Gra
ndN
EW
WS
-0.0
6
MFI
P
-0.0
7 Rap
ids
NE
WW
S
-0.1
7 **
Riv
ersi
deN
EW
WS
-0.1
6
MFI
P
***
NE
WW
S
Edu
catio
n-fi
rst p
rogr
ams
Job-
sear
ch-f
irst
pro
gram
s
NO
TE
S: I
n ea
ch s
tudy
, chi
ldre
n w
ere
sele
cted
for
incl
usio
n in
the
sam
ple
on th
e ba
sis
of th
eir
age
at r
ando
m a
ssig
nmen
t or
thei
r ag
e at
follo
w-u
p.T
he N
EW
WS
sam
ple
incl
udes
chi
ldre
n of
sin
gle
mot
hers
in th
e N
EW
WS
eval
uatio
n ag
ed 3
-5 a
t the
beg
inni
ng o
f th
e st
udy
who
se p
aren
ts w
ere
rand
omly
sel
ecte
dto
part
icip
ate
in th
e tw
o-ye
ar f
ollo
w-u
p su
rvey
(sa
mpl
e si
zes
for
educ
atio
n-fi
rst p
rogr
ams:
Atla
nta
= 1
,026
, Gra
nd R
apid
s =
421
, Riv
ersi
de =
578
; sam
ple
size
s fo
r jo
b-se
arch
-fir
stpr
ogra
ms:
Atla
nta
= 9
02, G
rand
Rap
ids
= 4
41, R
iver
side
= 6
94).
The
MFI
P sa
mpl
e in
clud
es c
hild
ren
of p
aren
ts in
the
MFI
P ev
alua
tion
aged
6-8
at t
he ti
me
of th
e th
ree-
year
fol
low
-up
surv
ey (
aged
app
roxi
mat
ely
3-5
atth
e tim
e of
rand
om a
ssig
nmen
t) w
hose
par
ents
wer
e lo
ng-t
erm
rec
ipie
nts
in u
rban
cou
ntie
s an
d un
derw
ent r
ando
m a
ssig
nmen
t bet
wee
n A
pril
1, 1
994,
and
Oct
ober
31,
199
4 (s
ampl
e si
ze f
orFu
ll M
FIP
= 2
86; s
ampl
e si
ze f
or M
FIP
Ince
ntiv
es O
nly
= 2
89).
The
SSP
sam
ple
incl
udes
chi
ldre
n of
sin
gle
pare
nts
in th
e SS
P ev
alua
tion
aged
6-8
at t
he ti
me
of th
e th
ree-
year
fol
low
-up
surv
ey (
aged
app
roxi
mat
ely
3-5
at r
ando
mas
sign
men
t) w
ho w
ere
livin
g in
the
hom
e at
ran
dom
ass
ignm
ent a
nd a
t the
tim
e of
the
thre
e-ye
ar f
ollo
w-u
p su
rvey
(sa
mpl
e si
ze =
1,3
18).
Stat
istic
al s
igni
fica
nce
leve
ls a
re in
dica
ted
as: *
= 1
0 pe
rcen
t; **
= 5
per
cent
; ***
= 1
per
cent
(tw
o-ta
iled
test
).In
NE
WW
S, h
ealth
was
mea
sure
d us
ing
a si
ngle
-ite
m m
easu
re th
at a
sked
par
ents
to r
ate
thei
r ch
ild's
gen
eral
hea
lth o
n a
five
-poi
nt s
cale
ran
ging
fro
m 1
("po
or")
to 5
("ex
celle
nt")
.In
MFI
P, h
ealth
was
ass
esse
d us
ing
a si
ngle
-ite
m m
easu
re th
at a
sked
par
ents
to r
ate
thei
r ch
ild's
hea
lth o
n a
five
-poi
nt s
cale
ran
ging
fro
m 1
("p
oor"
) to
5 ("
very
goo
d").
In S
SP, h
ealth
was
mea
sure
d us
ing
pare
nts'
res
pons
es to
fou
r qu
estio
ns a
bout
thei
r ch
ild's
hea
lth o
n a
scal
e ra
ngin
g fr
om 1
("f
alse
") to
5 (
"tru
e").
The
res
pons
es w
ere
aver
aged
acr
oss
the
four
item
s to
com
pute
a s
ingl
e sc
ore
for
each
chi
ld.
74
ticipants and/or the Atlanta programs' more supportive approach may account for their havingmore beneficial effects than the programs in the other two sites;However, the results also indi-cate that program site cannot have been the only factor influencing the impacts on children (forinstance, the Atlanta education-first program's impacts on children were not as positive as theAtlanta job-search-first program's impacts), but exploration of these factors as well as the sitedifferences lies outside the scope of this document.
Unlike the findings for programs with earnings supplements, these site differences sug-gest that programs with mandatory employment services do not necessarily have unidirectionaleffects on children. Depending on as yet unknown factors, these programs sometimes affectedchildren positively, negatively, or in most cases neutrally.
III. Effects on Child Care, Parents' Emotional Well-Being,and Parenting Behavior
As discussed earlier, programs with mandatory employment services increased employmentbut generally not income. How did children's daily lives change as their parents went to work?
Most of the programs with mandatory employment services increased young children'sexposure to child care, some of it formal care (for instance, that provided by child care centers).15Two of these programs also increased parents' feelings of time pressure (recorded as "feelingrushed"), but none of the programs significantly increased parents' feeling aggravated with theirchildren or parental depression.16 With respect to changes in the quality of parents' interactionswith their children, in two of the six programs parents reported having greater feelings of warmthtoward their children than parents in control group families (although one program reduced pa-rental warmth). The general lack of negative effects on parental emotional well-being and parent-ing behavior may explain why these children were not consistently affected by mandatory em-ployment services in an adverse way.
IV. Summary and Discussion of the Effects of Programs with MandatoryEmployment Services
It appears that mandatory employment services have few and mixed effects on children.The effects also appear to be specific to particular sites rather than linked to the presence ofmandatory employment services per se or to the fact that the program took an education-first ora job-search-first approach. Possible sources of explanation for these site differences in effectson children are differences in the local economic conditions, populations being served, andprogram implementation.
15 The two-year findings from the NEWWS Child Outcomes Study (McGroder et al., 2000) suggest that theseprograms did not consistently increase employment-related child care use for these samples of families with youngchildren at the time of the two-year survey. However, as indicated here, over the two-year period four of the pro-grams increased children's participation in any form of child care, and two of the programs increased use of formalchild care both irrespective of parental employment status.
I6Analyses conducted as part of the NEWWS Child Outcomes Study (McGroder et al., 2000) found that oneprogram reduced parental depression. Minor differences between the covariates included in the statistical modelsused there and here may account for the slight discrepancy between findings.
75-50-
Virtually all states have sanction policies that reduce families' welfare grants if parentsfail to comply with requirements to participate in employment-related activities. Should the find-ings reported in this chapter allay fears that children are harmed by mandatory participation re-quirements for parents receiving welfare? In some cases, but not always.
In the programs examined here, imposing a sanction for noncompliance with the partici-pation mandate entailed reducing the family's monthly welfare grant by the adult portion of thegrant and leaving the child portion unchanged. These sanctions known as partial family sanc-tions typically reduced the welfare grant by 15 percent to 20 percent. While 33 states currentlyhave similar partial sanctions in place as the first penalty that welfare recipients face for nonpar-ticipation, in only 15 states are such partial sanctions the maximum sanction imposed on fami-lies. The other states impose full family sanctions, eliminating all of the family's welfare grant.
In assessing what these sanctions mean to family income, it is important to consider thegrant level in the state in addition to differences in sanctions because grant levels differ substan-tially by state. A large sanction in a low-grant state can result in the same change in net incomeas a smaller sanction in a high-grant state. In South Carolina, for example, where the maximummonthly grant level for a family of three is about $200, the maximum sanction entails termina-tion of the entire welfare grant when recipients do not meet participation requirements. In Wash-ington where the maximum monthly grant level for a family of three is somewhat more than$500 in contrast, the maximum sanction entails elimination of 40 percent of the grant, orabout $200. In both states, a sanctioned family loses about $200 in monthly welfare benefits. InWashington, however, the result is a grant equivalent to 60 percent of the original amount, whilein South Carolina the result is no grant at all.
The studies examined in this chapter suggest that welfare policies that increase employ-ment by providing mandatory employment services but do not affect income are unlikely to causeconsistent patterns of harm or benefit to children. Nevertheless, programs that impose large sanc-tions on families in which parents fail to participate in required activities could lead to substantialloss of family income. It is unclear whether children would fare the same in such programs asthey did in the programs examined in this chapter.
1-51-
76
Chapter 4
Effects on Children of Programs with Time Limits
Time limits on welfare receipt have caused concern among observers who fear that manyparents will not be able to find work, leaving these parents and their children without the "safetynet" of welfare benefits. Others hope that the message conveyed by time limits will encourageparents to find work and reduce the negative effects on children of what they characterize as a"culture of dependency." At present the data bearing on how time limits affect parents and chil-dren are very limited. Moreover, there are no data at all on the effects of time limits alone as op-posed to time limits combined with other program features. In this chapter, we rely on the onlycompleted study that focused on the effects on children of a time-limited welfare program, a pro-gram that began prior to passage of PRWORA, the 1996 welfare reform law that placed a five-year lifetime limit on most families' receipt of federal cash welfare assistance. This was the Fam-ily Transition Program (FTP), a pilot that operated in Escambia County, Florida under waiversof AFDC rules from 1994 to 1998.'
I. Effects on Parents' Economic Outcomes
Time limits are intended to reduce families' long-term receipt of welfare and to increaseemployment among single parents receiving welfare; they are not intended to increase incomedirectly. Unlike the studies of the programs with earnings supplements and mandatory employ-ment services covered in Chapters 2 and 3, respectively, the FTP evaluation allows us to examinethe effects of time limits only when this feature is combined with other program features (includ-ing a small earnings supplement) rather than when implemented alone.
FTP increased employment and reduced welfare receipt, but these effectsdid not exceed those of the programs examined in earlier chapters.
Analyses conducted in a companion document2 indicate that FTP and another programcombining time limits with other approaches (Connecticut's Jobs First program) increased em-ployment. However, the effects on employment were generally no larger than those of programswith earnings supplements or of programs that provided only mandatory employment services.The time-limited programs also reduced welfare receipt, typically as families started reaching thetime limits.
FTP's time limits seem to have offset any income gains resulting from itsearnings supplement.
FTP's impacts on income were very small both before and after the first program groupmembers to undergo random assignment began to reach the time limits. Many observers haveworried that time limits may lead to large losses in family income, but the FTP findings suggestthat any such income loss owing to time limits is quite modest at least in the short term. Con-necticut's Jobs First program (for which data on children are not yet available) combines a time
'This section is based on a reanalysis of the data from the FTP evaluation (Bloom, Kemple, et al., 2000).2Bloom and Michalopoulos, 2001.
7 7-52-
limit with a generous earnings supplement. In the Jobs First evaluation, the income gains thatwere observed before families began to reach the time limit disappeared thereafter, when familieswere no longer eligible for the earnings supplement.3
Owing to its intensive case management and services, FTP was expensiverelative to programs that provided only mandatory employment services.
Florida implemented time limits in its welfare program cautiously, providing recipientswith intensive case management and services to help them find and keep jobs. These expenseswere not offset by the welfare savings generated by the program (in part because so many peoplein the control group left welfare anyway) making FTP, the net cost of which was nearly$8,000 per family over five years, expensive relative to programs that provided mandatory em-ployment services only. It is likely that other time-limited welfare programs would not be socostly, unless they offered the kinds of services and supports included in FTP.
II. Effects on Children
FTP's effects on the school achievement, behavior, and health of children aged 5-12 atthe end of the study period are presented in Figure 4.1.
FTP had few, and mixed, effects on children.
As Figure 4.1 illustrates, FTP did not significantly improve school achievement or reducebehavior problems.4 Although the program decreased positive behavior, it also improved health.
With regard to child care and family outcomes, FTP increased the use of child care. Theprogram had virtually no effect on aspects of the home environment. Among single parents, theprogram had no effect on the proportion of parents who married or experienced domestic violenceduring the study period. FTP also did not affect single parents' emotional well-being or parentingbehavior, except that it reduced parental supervision of elementary school-aged children.
Notably, the earnings supplement in FTP was very modest in size, much less generousthan the supplements offered in SSP and MFIP.5 Moreover, members of the FTP program staffdid not strongly emphasize the availability of the earnings supplement, nor did they suggest toparents that they prolong their use of welfare while working in order to increase their income.This approach stands in sharp contrast to those of MFIP, SSP, and New Hope, which were billedas make-work-pay programs with the earnings supplement as the centerpiece. Therefore, FTP'smore modest effects relative to the programs with earnings supplements and no time limits may
3Bloom, Melton, Michalopoulos, Scrivener, and Walter, 2000.'Interestingly, analyses conducted separately for younger children (aged 3-5) and older children (aged 6-8) re-
veal differences between the age groups in the program's effects on social behavior. Whereas the impacts were nega-tive for younger children, they were positive for older children. However, because these effects were not statisticallysignificant (although the differences between the effects were statistically significant), the findings are only sugges-tive of age specificity in FTP's effects. There were no differences between the two age groups in impacts onachievement or health.
5FTP's supplement was less generous primarily because the state's grant level was low rather than because theproportion of income that was disregarded was low.
78-53-
How Welfare and Work Policies Affect Children
Figure 4.1The Only Study of Time-Limited Welfare FoundFew and Mixed Effects on Children's Outcomes
0.30
0.25
0.20
0.150.09 0.09 *
0.10
0.05 0.01
0.00
-0.05
-0.10
-0.15 -0.11 *
Achievement Behavior Problems Positive Behavior Health
NOTES: The FTP sample includes children of single parents in the FTP evaluation aged 5-12 at the time of the four-year follow-upsurvey (aged approximately 1-8 at the time of random assignment) whose parents underwent random assignment between August1994 and February 1995 and participated in the four-year follow-up survey (sample size = 1,108).
Statistical significance levels are indicated as: * = 10 percent; ** = 5 percent; *** = 1 percent (two-tailed test).Achievement was assessed using a single-item measure that asked parents to rate their child's overall performance in
school. Responses range from I ("not well at all") to 5 ("very well").Behavior problems were measured using parents' responses to a 15-item externalizing subscale of the Behavioral
Problems Index, a subscale designed to assess the extent to which the child engaged in acting out and aggressive behaviors.Responses range from I ("not true") to 3 ("often true"). The responses to the 15 questions were summed to compute a single scorefor each child.
Positive behavior was measured using parents' responses to a subset of seven items from the Positive Behavior Scale thatassess positive aspects of children's behavior such as helpfulness and warmth. Responses range from 0 ("not at all like my child")to 10 ("completely like my child"). The responses to the seven questions were summed to capture a single score for each child.
Health was assessed using a single-item measure that asked parents to rate the child's health on a five-point scale rangingfrom I ("poor") to 5 ("very good").
79-54-
have as much to do with its less generous supplement and lack of a make-work-pay message aswith its time limits.
However, one might expect that an earnings supplement provided within the welfare sys-tem would have very different effects when combined with time limits, even if the supplementwere more generous than the one provided in FTP. Because their goals differ, time limits andearnings supplements may work at cross-purposes. Time limits encourage people to leave wel-fare quickly and save their remaining months of welfare eligibility for a period of crisis. Earningssupplements (when provided within the welfare system, as in the MFIP programs), in contrast,encourage families to continue to receive welfare benefits while they are working. Although sup-plements reduce the proportion of families relying solely on welfare, they typically increase theproportion of families receiving welfare because they lead more families to combine welfare andwork. Because the supplement comes from the welfare system, families are therefore likely touse up more months of their welfare eligibility if they are eligible to receive a supplement than ifa supplement were not available to them. Owing to the tension between time limits and earningssupplements, the effects on family income of a program that combines these two program fea-tures, although hard to predict, are likely to be smaller than those of programs that provide earn-ings supplements without imposing time limits.
What are the effects on children of a program with a generous earnings supplement andtime-limited welfare receipt? The interim results from the evaluation of Connecticut's Jobs Firstprogram (based on data collected 18 months after random assignment) suggest that any incomegains produced by such a program disappear as soon as time limits are instituted.6 Further resultson the effects of this program on children's outcomes (covering the 36 months after randomassignment) will be released late in 2001.
III. Summary and Discussion of the Effects of Programs with Time Limits
The limited evidence that is presently available on the effects on children of welfare timelimits suggests that these effects are few. However, state welfare programs with time limits willnot necessarily have similar effects on children as those of FTP because FTP included severalsafeguards to protect families from a loss of income owing to the time limits. First, some families(such as families in which parents were disabled or were responsible for caring for children under6 months old) were exempted from the study at the outset; if they had not been, these familiesmight have had a particularly difficult time being subject to time limits. Second, if a physicianfound a parent to be incapacitated (that is, unable to work), the months in which the parent wasincapacitated and the family received welfare were not counted toward the time limit during thestudy period. This second ground for exemption was invoked frequently (by 21 percent of thefamilies with enough months of welfare to exceed the time limit). Third, four-month extensionsof the time limit could be granted to parents who were deemed to have complied with the pro-gram's participation requirements but could not find work (this provision was rarely invoked).Finally, families could still receive the child portion of the grant after they reached the time limitif termination of welfare benefits might result in children's removal from the home (which was
6Bloom, Melton, et al., 2000.
80-55-
the case for nine out of the 237 families who reached the time limit). Together, these safeguardsmay have limited the amount of income that FTP families lost owing to the time limits. On thebasis of this evaluation alone, it is therefore impossible to conclude whether a time-limited pro-gram that, unlike FTP, results in a loss of family income would have the same neutral effects onchildren as FTP. It is also unclear what the effects on children of programs with time limits mightbe in the longer term that is, beyond the four-year period included in this study, which ex-tended only one to two years after families in FTP began reaching the time limits.
The findings from FTP suggest one additional important lesson about program cost: Pro-grams may need to spend additional money to improve children's well-being, but increasedspending may not be sufficient to achieve this goal. Families in FTP were offered an array of ser-vices to help them find and keep jobs and reduce their use of welfare. The FTP offices were alsowell staffed, with very low client-staff ratios. As a result, the cost of FTP per family relative tothe cost per control group family was high. The programs that included only mandatory employ-ment services examined in Chapter 3 cost much less, and some even saved the governmentmoney if the welfare savings these programs generated are taken into account.7 Yet both FTP andthe programs that provided only mandatory employment services led to few consistent effects onchildren, implying that spending more money does not necessarily improve children's lives. Thisis not to say that more expensive welfare reforms cannot help children. Indeed, the earnings sup-plement programs examined in Chapter 2, which were costly because they provided additionalincome to families without generating the same savings in cash benefits as programs that pro-vided only mandatory employment services, had generally beneficial or neutral effects on chil-dren. Policymakers may help children by providing income to families in which parents work.But spending that money in a way that does not result in an increase in family income, as in FTP
rather than in a way that increases family income, as in MFIP, New Hope, and SSP doesnot appear to bring the same benefits to children.
'See Bloom and Michalopoulos, 2001.
61-56-
Chapter 5
Effects of Welfare and Employment Programson Very Young Children and Adolescents
The findings analyzed in this monograph focus on the effects of welfare and employmentprograms on children who were preschool-aged or early school-aged at the time of their parents'random assignment and who were generally of elementary school age at the time of the follow-upassessments. At present, only a limited amount of information is available on the effects of suchprograms on very young children and adolescents. One might expect the effects to differ acrossage groups, perhaps especially at the ends of the age continuum. For instance, very young chil-dren may be more sensitive to separations from their mothers than are their older peers, and olderchildren are more likely than younger children to be left unsupervised and asked to take onhousehold responsibilities when their parents go to work. While it is beyond the scope of thismonograph to explore the programs' effects on very young children or adolescents, we nowbriefly discuss the effects on adolescents reported in the evaluations of SSP and FTP (the onlystudies covered here that examined data on adolescents): Recall that both programs increasedparental employment, but only SSP raised income more than modestly.
In the two studies that reported separate analyses of adolescents, someunfavorable effects on school and behavior outcomes were observed.
The findings from the SSP evaluation2 presented in Figure 5.1 suggest that the program hadnegative effects on adolescent children, although these results should be interpreted cautiously be-cause the outcomes were assessed for only about two-thirds of the adolescents whose parents werein the study. On average, parents in the program group reported lower average achievement for ado-lescents than parents in the control group, and adolescent children of parents in the program groupwere more likely than their control group counterparts to report performing below average in school(although the impact on the adolescent-reported measure was not statistically significant).3 Withrespect to children's behavior, however, SSP's unfavorable effects were much more consistent. Theprogram increased adolescent children's behavior problems in school as reported by parents as wellas smoking and weekly alcohol use as reported by adolescents. The oldest adolescent children ofparents in the program group were also more likely than their control group counterparts to reportengaging in minor delinquent activity (such as staying out later than their parents allowed; notshown in figure). It should be noted that this was the same program that had such positive effects onchildren who were of preschool or early school age at the time of their parents' random assignment(and were aged 6-11 at the time of the three-year follow-up assessment).
Figure 5.2 presents FTP's effects on adolescent children.4 As the figure shows, adoles-cents with parents in the program group reported doing slightly worse in school and were more
'Morris and Michalopoulos, 2000; Bloom, Kemple, et al., 2000.2Morris and Michalopoulos, 2000.3However, these findings were not corroborated by children's scores on a math skills test, on which no differ-
ence between the program and control groups was found (not shown in figure).4Bloom, Kemple, et al., 2000.
82-57-
How Welfare and Work Policies Affect Children
Figure 5.1SSP Increased Adolescents' Behavior Problems
0.30
0.25
0.20
0.15
°Parents' reports Adolescents' reports
0.20 ***
0.11 *0.09 *
0.10
0.05
0.00
-0.05
-0.05-0.10
-0.09-0.15 -0.11 *
Achievement School Behavior Smoking Drinking HealthProblems
NOTES: The SSP sample includes children of single parents in the SSP evaluation aged 12-18 at the time of the three-yearfollow-up survey (aged approximately 9-15 at random assignment) who were living in the home at the time of randomassignment and at the time of the three-year follow-up survey (sample size = 1,417).
Statistical significance levels are indicated as: * = 10 percent; ** = 5 percent; *** = 1 percent (two-tailed test).Achievement was measured using parents' and children's responses to questions about the child's functioning in
three academic subjects. The responses, which were expressed on a five-point scale ranging from 1 ("not very well") to 5("very well"), were averaged across the three subjects to compute a single score for each child.
School behavior problems were assessed using parents' responses to a single-item measure that asked how often inthe past school year they were contacted by the school about their child's behavior problems in school.Responses range from 1("never contacted or contacted once") to 3 ("contacted four times or more").
Smoking was assessed using children's responses to a single-item measure that asked whether or not they currentlysmoked.
Drinking was assessed using children's responses to a single-item measure about their frequency of alcohol use in theprior six months. Responses range from "never" to "every day. " If the child reported using alcoholat least weekly, theresponse was coded as 0; otherwise it was coded as 1.
Health was measured using parents' responses to four items about their child's health on a scale ranging from 1("false") to 5 ("true"). The responses were averaged across the four items to compute a single score for each child.
-58- 3
How Welfare and Work Policies Affect Children
Figure 5.2FTP Had Two Unfavorable Effects on Adolescents
0.25
0.20 0.17 **
0.15
0.1 0.10
tom 0.05 0.01
14.co) 0.00
4g -0.05 -0.03
-0.10
-0.15-0.14 *
-0.20
Achievement Ever Suspended Ever Arrested Ever Had a Baby
NOTES: The FTP sample includes children of single parents in the FTP evaluation aged 13-17 at the time of the four-yearfollow-up survey (aged approximately 9-13 at random assignment) whose parents underwent random assignment betweenAugust 1994 and February 1995 and participated in the four-year follow-up survey (sample size = 741).
Statistical significance levels are indicated as: * = 10 percent; ** = 5 percent; *** = 1 percent (two-tailed test).Achievement was assessed using a single-item measure that asked parents to rate their child's overall performance in
school. Responses range from 1 ("not well at all") to 5 ("very well").School suspension was assessed using a single-item measure that asked parents if their child had ever been suspended
from school since random assignment.Police involvement was assessed using a single-item measure that asked parents if their child had been arrested since
random assignment for any offense other than a minor traffic violation.Fertility was assessed using a single-item measure that asked parents if their child had fathered a baby or had a baby
since random assignment.
4-59-
likely to be suspended from school during the study period than their counterparts in the controlgroup. However, no significant effects were found between the groups with respect to whetherthey were ever arrested or became pregnant during the study period. While FTP had fewer effectson adolescents than SSP, together the findings suggest that programs that move parents fromwelfare into employment may have some negative effects on adolescent children.
The two programs were very different in their policy features: Whereas SSP offered onlyan earnings supplement, FTP combined a small earnings supplement with time limits and manda-tory employment services. Moreover, the programs' effects on economic outcomes for parentsdiffered as well: FTP's effects on income were smaller than those of SSP. Given that it is througheconomic outcomes that children are most likely to be affected by features of welfare and em-ployment programs, the fact that the two programs had negative or neutral effects on adolescentsis noteworthy. Nevertheless, based on these limited data it remains unclear whether such impactson outcomes for adolescents are likely to be found in other program approaches as well. Futurework will focus on adolescent children to see if SSP' s and FTP's effects are consistent with thoseof the other programs examined in this monograph.
-60-
P 1-
Chapter 6
Conclusions and Policy Implications
The research synthesized in the previous chapters supports the following conclusionsabout the effects on children of the 11 welfare and employment programs examined in thismonograph (see Figure 6.1 for a summary of their effects on school achievement).
The programs that aimed to promote parental employment through earn-ings supplements had positive impacts on children's well-being.
The positive impacts of the four programs with earnings supplements examined here werelargest for children in long-term recipient families and for elementary school-aged children.
Although the earnings supplement programs increased employment and income, they leftmany families disadvantaged. Substantial fractions of children whose families were in even themost generous programs were not progressing normally in school, lived in families that were stillpoor, and had parents who were depressed.
The programs with mandatory employment services, all of which boostedparental employment without increasing income, had few and mixedeffects on children.
These six programs had relatively few noteworthy effects on children. When im-pacts were found, they were about equally likely to be positive as negative. Whether therewere impacts appeared to be more closely associated with particular program sites thanwith program characteristics such as participation mandates.
The one program with time limits, which led to an increase in parentalemployment and a modest increase in income, produced few noteworthyimpacts on children, and the impacts found were mixed.
Our knowledge base is smallest with regard to the impacts of time limits becausethe only program with time limits combined them with mandatory employment servicesand a small earnings supplement. The program's few impacts on children were mixed:Health improved, but positive social behavior decreased.
In the two studies that examined adolescent children, the programs ap-peared to be less beneficial for adolescents than for children in middlechildhood.
The five studies examined in this monograph gathered much more information about thewell-being of elementary school-aged children than they did about very young children or ado-lescents. In the two studies that examined adolescents, however, the findings suggest that par-ents' transition from welfare to work may decrease adolescents' school achievement and increasetheir behavior problems, perhaps by lessening parents' ability to maintain communication withand monitor the behavior of their adolescent children.
-61-
How
Wel
fare
and
Wor
k Po
licie
s A
ffec
t Chi
ldre
n
Figu
re 6
.1Su
mm
ary
of A
ll 11
Pro
gram
s' I
mpa
cts
on C
hild
ren'
s Sc
hool
Ach
ieve
men
t
0.35
0.30
Prog
ram
s w
ith e
arni
ngs
supp
lem
ents
Prog
ram
s w
ith m
anda
tory
em
ploy
men
t ser
vice
s on
lyPr
ogra
m w
ithtim
e lim
its
0.25
**
0.25
0.19
***
0.20
0.14
*0.
14 *
0.15
*0.
150.
100.
090.
10
0.05
0.03
0.03
0.03
0.00
-0.0
5-0
.04
-0.1
0
MFI
PSS
PN
ew H
ope
Full
MFI
PA
tlant
aG
rand
Rap
ids
Riv
ersi
deA
tlant
aG
rand
Rap
ids
Riv
ersi
deFT
PIn
cent
ives
NE
WW
SN
EW
WS
NE
WW
SN
EW
WS
NE
WW
SN
EW
WS
Onl
yE
duca
tion-
firs
t pro
gram
sJo
b-se
arch
-fir
st p
rogr
ams
NO
TE
S: I
n ea
ch s
tudy
, chi
ldre
n w
ere
sele
cted
for
incl
usio
n in
the
sam
ple
on th
e ba
sis
of th
eir
age
at r
ando
m a
ssig
nmen
t or
thei
r ag
e at
follo
w-u
p.T
he p
rogr
ams
with
ear
ning
s su
pple
men
ts a
re th
e fo
ur p
rogr
ams
in th
e M
FIP,
SSP
, and
New
Hop
e ev
alua
tions
. The
MFI
P sa
mpl
e in
clud
es c
hild
ren
of p
aren
ts in
the
MFI
P ev
alua
tion
aged
5-1
2 at
the
time
of th
e th
ree-
year
follo
w-u
p su
rvey
(ag
ed a
ppro
xim
atel
y 2-
9 at
the
time
of r
ando
m a
ssig
nmen
t) w
hose
par
ents
wer
e lo
ng-t
erm
rec
ipie
nts
in u
rban
cou
ntie
s an
d un
derw
ent r
ando
m a
ssig
nmen
t bet
wee
n A
pril
1, 1
994,
and
Oct
ober
31,
199
4 (s
ampl
e si
zefo
r Fu
ll M
FIP
= 5
87; s
ampl
e si
ze f
or M
FIP
Ince
ntiv
es O
nly
= 5
73).
The
SSP
sam
ple
incl
udes
chi
ldre
n of
sin
gle
pare
nts
in th
e SS
P ev
alua
tion
aged
6-1
1 at
the
time
of th
e th
ree-
year
fol
low
-up
surv
ey (
aged
app
roxi
mat
ely
3-8
atra
ndom
ass
ignm
ent)
who
wer
e liv
ing
in th
e ho
me
at th
e tim
e of
ran
dom
ass
ignm
ent a
nd a
t the
tim
e of
the
thre
e-ye
ar f
ollo
w-u
p su
rvey
(sa
mpl
e si
ze=
2,1
58).
The
New
Hop
e sa
mpl
e in
clud
es c
hild
ren
of th
e si
ngle
par
ents
in th
e N
ewH
ope
eval
uatio
n w
ho w
ere
aged
1-1
0 at
ran
dom
ass
ignm
ent a
nd w
hose
par
ents
par
ticip
ated
in th
e tw
o-ye
ar f
ollo
w-u
p su
rvey
(sa
mpl
e si
ze =
832
).T
he p
rogr
ams
with
man
dato
ry e
mpl
oym
ent s
ervi
ces
are
the
six
prog
ram
s in
the
NE
WW
S ev
alua
tion.
The
NE
WW
S sa
mpl
e in
clud
es c
hild
ren
of s
ingl
e m
othe
rs in
the
NE
WW
Sev
alua
tion
aged
3-5
at t
he b
egin
ning
of
the
stud
y w
hose
par
ents
wer
e ra
ndom
ly s
elec
ted
to p
artic
ipat
e in
the
two-
year
fol
low
-up
surv
ey (
sam
ple
size
s fo
r ed
ucat
ion-
firs
t pro
gram
s: A
tlant
a =
1,0
26, G
rand
Rap
ids
= 4
21,
Riv
ersi
de =
578
; sam
ple
size
s fo
r jo
b-se
arch
-fir
stpr
ogra
ms:
Atla
nta
= 9
02, G
rand
Rap
ids
= 4
41, R
iver
side
= 6
94).
The
pro
gram
with
tim
e lim
its is
the
prog
ram
in th
e FT
P ev
alua
tion.
The
FT
P sa
mpl
e in
clud
es c
hild
ren
of s
ingl
e pa
rent
s in
the
FTP
eval
uatio
n ag
ed5-
12 a
t the
tim
e of
the
four
-yea
r fo
llow
-up
surv
ey (
aged
app
roxi
mat
ely
1-8
at th
e tim
e of
ran
dom
ass
ignm
ent)
who
se p
aren
ts u
nder
wen
t ran
dom
ass
ignm
ent b
etw
een
Aug
ust 1
994
and
Febr
uary
199
5 an
d pa
rtic
ipat
ed in
the
four
-yea
r fo
llow
-up
surv
ey (
sam
ple
size
= 1
,108
).St
atis
tical
sig
nifi
canc
e le
vels
are
indi
cate
d as
:=
10
perc
ent;
= 5
per
cent
; **
= 1
per
cent
(tw
o-ta
iled
test
).In
MFI
P, a
chie
vem
ent w
as a
sses
sed
usin
g a
sing
le-i
tem
mea
sure
that
ask
ed p
aren
ts to
rat
e th
eir
child
's o
vera
ll pe
rfor
man
ce in
sch
ool o
n a
scal
e ra
ngin
g fr
om1
("no
t wel
l at a
ll")
to 5
("v
ery
wel
l").
In S
SP, a
chie
vem
ent w
as m
easu
red
usin
g a
26-
to 3
4-ite
m m
ath
skill
s te
st a
nd e
xpre
ssed
in te
rms
of th
e pr
opor
tion
of it
ems
answ
ered
cor
rect
ly. P
aren
ts'
asse
ssm
ents
of
achi
evem
ent w
ere
mea
sure
d us
ing
thei
r ra
tings
of
thei
rch
ild's
fun
ctio
ning
in th
ree
acad
emic
sub
ject
s on
a f
ive-
poin
t sca
le r
angi
ng f
rom
1 (
"not
ver
y w
ell"
) to
5 (
"ver
y w
ell"
). T
he r
atin
gs w
ere
aver
aged
acr
oss
the
thre
e ac
adem
ic s
ubje
cts
to c
ompu
te a
sin
gle
scor
e fo
r ea
ch c
hild
.In
New
Hop
e, te
ache
rs' r
epor
ts o
f ac
hiev
emen
t wer
e m
easu
red
usin
g th
e 10
-ite
m A
cade
mic
Sub
scal
e fr
om th
e So
cial
Ski
lls R
atin
g Sy
stem
, whi
ch a
sked
teac
hers
to r
ate
the
child
's s
kills
rel
ativ
e to
thos
e of
oth
er c
hild
ren
inar
eas
such
as
mat
h, r
eadi
ng, a
nd o
ral c
omm
unic
atio
n on
a f
ive-
poin
t sca
le r
angi
ng f
rom
1 (
"bot
tom
10
perc
ent"
) to
5 (
"top
10
perc
ent"
). T
he r
espo
nses
wer
e av
erag
ed a
cros
s th
e 10
item
s to
com
pute
a s
ingl
e sc
ore
for
each
chi
ld.
Pare
nts'
ass
essm
ents
of
achi
evem
ent w
ere
mea
sure
d us
ing
a si
ngle
-ite
m m
easu
re th
at a
sked
par
ents
to r
ate
thei
r ch
ild's
sch
ool p
erfo
rman
ce, b
ased
on
past
rep
ort c
ards
oro
ther
sou
rces
, on
a fi
ve-p
oint
sca
le r
angi
ng f
rom
1 (
"not
at a
llw
ell"
) to
5 (
"ver
y w
ell"
).In
NE
WW
S, a
chie
vem
ent w
as m
easu
red
usin
g ch
ildre
n's
stan
dard
sco
res
on th
e B
rack
en S
choo
l Rea
dine
ss C
ompo
site
test
, whi
ch a
sses
ses
know
ledg
e of
col
ors,
lette
rs, n
umbe
rs/c
ount
ing,
com
pari
sons
, and
shap
es.
In F
TP,
ach
ieve
men
t was
ass
esse
d us
ing
a si
ngle
-ite
m m
easu
re th
at a
sked
par
ents
to r
ate
thei
r ch
ild's
ove
rall
perf
orm
ance
in s
choo
l. R
espo
nses
ran
ge f
rom
I (
"not
wel
l at a
ll")
to 5
("v
ery
wel
l").
BE
ST C
OPY
AV
AIL
AB
LE
I. Policy Implications
What do the five studies covered in this document suggest will be the likely impacts onchildren of current and future welfare and employment policies? The studies' use of random as-signment designs lends weight to the conclusions drawn here about the overall impacts of theseprograms on children's well-being. Still, there are limits to how far the findings can be general-ized in the current policy environment. The programs themselves do not represent the full rangeof program features being implemented by states since the replacement of AFDC by TANF, norwere they tested in the full range of macroeconomic conditions good and bad that statescurrently face or are likely to face in the next decade. Moreover, our knowledge about the effectsof time-limited welfare is limited to a single study. Finally, because all the studies followed chil-dren for only two to four years, we do not yet know the effects of these programs on children'seducational attainment or children's expectations regarding work and childbearing in the longerterm. Bearing these caveats in mind, we now draw some policy implications suggested by theseresults.
Increases in income as well as parental employment may underlie somewelfare and employment programs' beneficial effects on children's devel-opment.
So much rhetoric and so many of the provisions for welfare reform have been focused onparental employment and welfare receipt that it is easy to lose sight of the fact that changes inparents' and families' circumstances can affect the development and well-being of children. Inparticular, we found that programs that provided earnings supplements had consistently positiveimpacts on children's achievement. Although the study designs do not enable us to identify pre-cisely what about the programs was responsible for these effects, thanks to random assignmentwe can be very confident in attributing the effects to some aspect or aspects of the programs.While it remains unclear whether the increases in income alone or the combination of increasesin income and employment together produced the positive effects on children, policymakers needto be aware that programs that supplement parents' earnings can have important positive effectson children. These findings suggest that earnings supplementation policies such as the EIC andchild care subsidies may be important for children as well.
States often view reductions in their TANF caseloads as indicators of successful policy.The research findings synthesized here suggest that they should be equally attentive to their pro-gress in reducing child poverty.
Earnings supplements may serve as another tool for state and federal governments to en-hance school achievement among low-income children. Past research indicates that some early-education programs promote the school achievement of preschool-aged children. Our resultssuggest that certain kinds of welfare and employment programs can have similar effects, particu-larly on children of long-term welfare recipients.
Raising employment without increasing income may not be sufficient toboost the healthy development of children in low-income families.
-63- Z;9
The findings presented in this monograph run counter to the prediction that increases inemployment alone would by enhancing parents' self-esteem, making family life more struc-tured, and giving children positive role models promote children's well-being, at least in theshort term. Although positive impacts on children's achievement and behavior were found, theseeffects were chiefly limited to the programs with earnings supplements; the effects on children ofprograms that included only mandatory employment services were few and mixed. The singlestudy of a program with time limits also revealed few effects on children and no consistent pat-tern of benefit or harm. Again, the programs in the five studies covered here do not reflect thefull range of current state TANF programs. Programs that impose stronger family sanctions fornoncompliance with participation mandates and/or fewer safeguards for families who reach thetime limit may have more harmful effects on children. Furthermore, a different pattern of effectsmay be observed for some subgroups of families and children than for the samples examinedhere.
Working parents may need help in their efforts to provide emotionalsupport and supervision to their children well into their children's ado-lescent years.
A given program may have quite different impacts on children of different ages, and ado-lescents in particular may be at risk when their parents are in programs that boost employment.After-school programs and youth development programs are potential avenues for promotingadolescents' positive development, as are efforts to strengthen neighborhoods and communities.1
Increases in government spending may be necessary for improving chil-dren's outcomes but are not sufficient.
A comparison of findings for the 11 programs examined in this monograph reveals a criti-cal tradeoff: Mandatory services by themselves have few effects on children and can save thegovernment money; earnings supplements can benefit children, but they are more costly. Pro-grams that increase family income and improve children's well-being may require increasedspending. However, as the results from the study of the single time-limited welfare program indi-cated, spending is not sufficient to improve outcomes for children: Despite being expensive, thisprogram had only modest effects on family income and had few and mixed effects on children. Inother words, increased funding can be deployed in ways that are more or less helpful to chil-dren's development, and increasing family income along with employment appears to be an im-portant component of that package.
II. Further Research from the Next Generation Project
In this document, we examined how welfare and employment policies targeted at low-income families can affect children. Analyses are currently under way to advance our understand-ing of some of the pathways through which these effects occurred. The analyses examine suchintervening mechanisms as family income, amount and type of parental employment, and use andtype of child care. Specifically, the dynamics of income change, including income growth and
'Lamer, Zippiroli, and Behrman, 1999.
-64-9 0
loss, are being tested as possible influences on children; parental transitions into and out of em-ployment, job stability and tenure, hours worked, hourly wages, and wage growth are being ana-lyzed in relation to children's well-being; and the effects of welfare and employment policies onchild care use and the relation between the amount and type of child care used and children's de-velopment at different ages are being explored.
This research synthesis represents a first step in the Next Generation project. The findingsfrom the project's ongoing work will be summarized and posted on the Next Generation Website and will be released in future reports.
Please check http://www.mdrc.org/NextGeneration for the latest information.
-65-
Appendix
Outcome Levels and ImpactsUnderlying the Effect Sizes Presented in
This Monograph
v,2
Listed below are descriptions of the measures of children's outcomes that are presented in Ap-pendix Tables 1-8. Each table presents the average outcomes in the program and control groupsseparately.
1. Measures Examined in the Earnings Supplement Programs (Tables 1-4)
Achievement
In MFIP, achievement was assessed using a single-item measure that asked parents torate their child's overall performance in school on a scale ranging from 1 ("not well atall") to 5 ("very well").
In SSP, achievement was measured using a 26- to 34-item math skills test and expressedin terms of the proportion of items answered correctly. Parents' assessments ofachievement were measured using their ratings of their child's functioning in three aca-demic subjects on a five-point scale ranging from 1 ("not very well") to 5 ("very well").The ratings were averaged across the three academic subjects to compute a single scorefor each child.
In New Hope, teachers' reports of achievement were measured using the 10-item Aca-demic Subscale from the Social Skills Rating System, which asked teachers to rate thechild's skills relative to those of other children in areas such as math, reading, and oralcommunication on a five-point scale ranging from 1 ("bottom 10 percent") to 5 ("top 10percent"). The responses were averaged across the 10 items to compute a single scorefor each child. Parents' assessments of achievement were measured using a single-itemmeasure that asked parents to rate their child's school performance, based on past reportcards or other sources, on a five-point scale ranging from 1 ("not at all well") to 5 ("verywell").
Behavior problems
In MEP, behavior problems were measured using parents' responses to a 12-item ex-ternalizing subscale of the Behavioral Problems Index that assesses aggressive behav-iors such as bullying and cheating. Responses range from 0 ("not true") to 2 ("verytrue"). The responses to the 12 questions were summed to compute a single score foreach child.
In SSP, behavior problems were measured using a four-item externalizing subscale thatasked parents to assess their child's acting out and aggressive behaviors on a three-pointscale ranging from 1 ("never") to 3 ("often"). The responses were averaged across thefour items to compute a single score for each child.
In New Hope, behavior problems were measured using a six-item externalizing subscaleof the Problem Behavior Scale from the Social Skills Rating System that asked parentsand teachers about the child's aggressive behavior and how often the child needed to bedisciplined for misbehavior on a five-point scale ranging from 1 ("never") to 5 ("all the
-68- X13
time"). The responses were averaged across the six items to compute a single score foreach child. ,
Positive behavior
In MFIP, positive behavior was measured with the 25-item Positive Behavior Scale,which included three subscales: compliance, social competence, and autonomy. Parentsresponded to each item on an 11-point scale ranging from 0 ("not at all like my child")to 10 ("completely like my child"). The responses to the 25 questions were summed tocompute a single score for each child.
In SSP, positive behavior was measured using the five-item Positive Social Behaviorsubscale, which asked parents to assess their child's prosocial interactions with peers ona scale ranging from 1 ("never") to 3 ("often"). The responses were averaged across thefive items to compute a single score for each child.
In New Hope, the child's positive behavior was measured using the 25-item PositiveBehavior Scale, which included three subscales: compliance, social competence, andautonomy. Parents and teachers responded to each item on a five-point scale rangingfrom 1 ("never") to 5 ("all of the time"). The responses were averaged across the 25items to compute a single score for each child.
Health
In MFIP, health was assessed using a single-item measure that asked parents to rate theirchild's health on a five-point scale ranging from 1 ("poor") to 5 ("very good").
In SSP, health was measured using parents' responses to four questions about theirchild's health on a scale ranging from 1 ("false") to 5 ("true"). The responses were aver-aged across the four items to compute a single score for each child.
2. Measures Examined in the Programs with Mandatory EmploymentServices (Tables 5 and 6)
Achievement was measured using children's standard scores on the Bracken SchoolReadiness Composite test, which assesses knowledge of colors, letters, num-bers/counting, comparisons, and shapes.
Behavior problems were measured using parents' responses to a 12-item externalizingsubscale of the Behavioral Problems Index, which assesses aggressive behaviors such asbullying and cheating. Responses range from 0 ("not true") to 2 ("very true"). The re-sponses to the 12 questions were summed to compute a single score for each child.
Positive behavior was measured using parents' responses to a 7-item scale assessing theextent to which children get along with peers. Responses range from 0 ("not true") to 2("very true"). The responses to the 7 items were summed to compute a single score foreach child.
94-69-
Health was measured using a single-item measure that asked parents to rate their child'sgeneral health on a five-point scale ranging from 1 ("poor") to 5 ("excellent").
3. Measures Examined in the Time-Limited Welfare Program (Table 7)
Achievement was assessed using a single-item measure that asked parents to rate theirchild's overall performance in school. Responses range from 1 ("not well at all") to 5("very well").
Behavior problems were measured using parents' responses to a 15-item externalizingsubscale of the Behavi9ral Problems Index, a subscale designed to assess the extent towhich the child engaged in acting out and aggressive behaviors. Responses rangelrom 1("not true") to 3 ("often true"). The responses to the 15 questions were summed to com-pute a single score for each child.
Positive behavior was measured using parents' responses to a subset of seven itemsfrom the Positive Behavior Scale that assess positive aspects of children's behavior suchas helpfulness and warmth. Responses range from 0 ("not at all like my child") to 10("completely like my child"). The responses to the seven questions were summed tocapture a single score for each child.
Health was assessed using a single-item measure that asked parents to rate the child'shealth on a five-point scale ranging from 1 ("poor") to 5 ("very good").
4. Measures Examined in the Two Programs That Assessed Adolescents(Table 8)
SSP
Achievement was measured using parents' and children's responses to questions aboutthe child's functioning in three academic subjects. The responses, which were expressedon a five-point scale ranging from 1 ("not very well") to 5 ("very well"), were averagedacross the three subjects to compute a single score for each child.
School behavior problems were assessed using parents' responses to a single-itemmeasure that asked how often in the past school year they were contacted by the schoolabout their child's behavior problems in school. Responses range from 1 ("never con-tacted or contacted once") to 3 ("contacted four times or more").
Smoking was assessed using children's responses to a single-item measure that askedwhether or not they currently smoked.
Drinking was assessed using children's responses to a single-item measure about theirfrequency of alcohol use in the prior six months. Responses range from "never" to"every day." If the child reported using alcohol at least weekly, the response was codedas 1; otherwise it was coded as 0.
-70- S5
Health was measured using parents' responses to four items about their child's health ona scale ranging from 1 ("false") to 5 ("true"). The responses were averaged across thefour items to compute a single score for each child.
FTP
Achievement was assessed using a single-item measure that asked parents to rate theirchild's overall performance in school. Responses range from 1 ("not well at all") to 5("very well").
School suspension was assessed using a single-item measure that asked parents if theirchild had ever been suspended from school since random assignment.
Police involvement was assessed using a single-item measure that asked parents if theirchild had been arrested since random assignment for any offense other than a minor traf-fic violation.
Fertility was assessed using a single-item measure that asked parents if their child hadfathered a baby or had a baby since random assignment.
C 6
-71-
How Welfare and Work Policies Affect Children
Table 1
Earnings Supplement Programs
Impacts for Children Who Were Preschool-Aged orEarly Elementary School-Aged at Random Assignment
Outcome
ProgramGroup
Average
ControlGroup
Average ImpactEffect Size
of Impact
AchievementMFIP Incentives Only
Parent report 4.11 3.96 0.16 * 0.14SSP
Parent report 3.71 3.61 0.10 ** 0.11Math skills test 0.56 0.52 0.04 ** 0.14
New HopeParent report 3.99 3.90 0.09 0.08Teacher report 3.33 3.09 0.24 ** 0.25
Full MFIPParent report 4.13 3.96 0.17 * 0.15
Externalizing behavior problemsMFIP Incentives Only
Parent report 5.21 6.02 -0.81 * -0.15SSP
Parent report 1.25 1.25 0.00 -0.01New Hope
Parent report 2.57 2.58 -0.01 -0.01Teacher report 2.12 2.20 -0.08 -0.09
Full MFIPParent report 5.12 6.02 -0.91 ** -0.17
Positive behaviorMFIP Incentives Only
Parent report 200.63 193.70 6.93 ** 0.18SSP
Parent report 2.58 2.59 -0.01 -0.02New Hope
Parent report 3.95 3.96 0.00 -0.01Teacher report 3.65 3.51 0.15 ** 0.21
Full MFIPParent report 194.20 193.70 0.50 0.01
HealthMFIP Incentives Only
Parent report 4.27 4.21 0.06 0.06SSP
Parent report 4.11 4.02 0.09 ** 0.11Full MFIP
Parent report 4.11 4.21 -0.09 -0.09
(continued)
37
Table 1 (continued)
NOTES: In each study, children were selected for inclusion in the sample on the basis of their age atrandom assignment or their age at follow-up.
The MFIP sample includes children of parents in the MFIP evaluation aged 5-12 at the timeof the three-year follow-up survey (aged approximately 2-9 at the time of random assignment) whoseparents were long-term recipients in urban counties and underwent random assignment between April1, 1994, and October 31, 1994 (sample size for Full MFIP = 587; sample size for MFIP IncentivesOnly = 573).
The SSP sample includes children of single parents in the SSP evaluation aged 6-11 at thetime of the three-year follow-up survey (aged approximately 3-8 at random assignment) who wereliving in the home at the time of random assignment and at the time of the three-year follow-up survey(sample size = 2,158).
The New Hope sample includes children of the single parents in the New Hope evaluationwho were aged 1-10 at random assignment and whose parents participated in the two-year follow-upsurvey (sample size = 832).
Because of rounding, the impacts shown may not exactly equal the differences between theprogram and control group averages.
Statistical significance levels are indicated as: * = 10 percent; ** = 5 percent; *** = 1percent (two-tailed test).
-73- 8
How Welfare and Work Policies Affect Children
Table 2
Earnings Supplement Programs
Impacts for Children of Long-Term Recipients Who Were Preschool-Aged orEarly Elementary School-Aged at Random Assignment
Outcome
ProgramGroup
Average
AchievementMFIP Incentives Only
Parent report 4.11SSP
Parent report 3.71Math skills test 0.55
New HopeParent report 3.99Teacher report 3.37
Full MFIPParent report 4.13
Externalizing behavior problemsMFIP Incentives Only
Parent report 5.21
SSPParent report 1.25
New HopeParent report 2.62Teacher report 2.10
Full MFIPParent report 5.12
Positive behaviorMFIP Incentives Only
Parent report 200.63SSP
Parent report 2.59New Hope
Parent report 3.95Teacher report 3.66
Full MFIPParent report 194.20
HealthMFIP Incentives Only
Parent report 4.27SSP
Parent report 4.12Full MFIP
Parent report 4.11
=74-
ControlGroup
Average ImpactEffect Size
of Impact
3.96 0.16 * 0.14
3.61 0.11 ** 0.11
0.53 0.02 0.09
3.77 0.22 * 0.202.99 0.38 *** 0.39
3.96 0.17 * 0.15
6.02 -0.81 * -0.15
1.25 0.00 0.01
2.61 0.01 0.012.18 -0.08 -0.10
6.02 -0.91 ** -0.17
193.70 6.93 ** 0.18
2.59 -0.01 -0.02
3.92 0.04 0.083.50 0.16 * 0.23
193.70 0.50 0.01
4.21 0.06 0.06
4.00 0.12 *** 0.14
4.21 -0.09 -0.09
(continued)
S 9
Table 2 (continued)
NOTES: In each study, children were selected for inclusion in the sample on the basis of their age atrandom assignment or their age at follow-up.
The MFIP sample includes children of parents in the MFIP evaluation aged 5-12 at the timeof the three-year follow-up survey (aged approximately 2-9 at the time of random assignment) whoseparents were long-term recipients in urban counties and underwent random assignment between April1, 1994, and October 31, 1994 (sample size for Full MFIP = 587; sample size for MFIP IncentivesOnly = 573).
The SSP sample includes children of parents in the SSP evaluation aged 6-11 at the time ofthe three-year follow-up survey (aged approximately 3-8 at random assignment) who were living inthe home at the time of random assignment and at the time of the three-year follow-up survey andwhose parents had at least two years of welfare receipt prior to random assignment (sample size =2,015).
The New Hope sample includes children of the single parents in the New Hope evaluationwho were aged 1-10 at random assignment and whose parents had at least two years of welfare receiptprior to random assignment and participated in the two-year follow-up survey (sample size = 508).
Because of rounding, the impacts shown may not exactly equal the differences between theprogram and control group averages.
Statistical significance levels are indicated as: * = 10 percent; ** = 5 percent; *** = 1percent (two-tailed test).
0-75-
How Welfare and Work Policies Affect Children
Table 3
Earnings Supplement Programs
Impacts for Children Who Were Preschool-Aged or Early Elementary School-Agedat Random Assignment, by Children's Age
Outcome
ProgramGroup
Average
ControlGroup
Average Impact
DifferenceBetween
Effect Size Subgroup
of Impact Impacts
AchievementMFIP Incentives Only
Children aged 3-5
Parent report 4.13 4.05 0.08 0.07SSP
Parent report 3.74 3.60 0.14 ** 0.15Math skills test 0.52 0.46 0.06 ** 0.22
New HopeParent report 4.16 4.15 0.01 0.01
Teacher report 3.49 3.18 0.31 ** 0.31Full MFIP
Parent report 4.21 4.05 0.16 0.15
Externalizing behavior problemsMFIP Incentives Only
Parent report 5.22 5.19 0.04 0.01SSP
Parent report 1.24 1.24 0.00 -0.01New Hope
Parent report 2.55 2.54 0.00 0.01Teacher report 1.98 2.15 -0.17 -0.19
Full MFIPParent report 4.95 5.19 -0.23 -0.05
Positive behaviorMFIP Incentives Only
Parent report 203.42 198.06 5.36 0.14SSP
Parent report 2.57 2.60 -0.02 -0.05New Hope
Parent report 3.93 3.95 -0.02 -0.04Teacher report 3.75 3.61 0.14 0.20
Full MFIPParent report 193.01 198.06 -5.05 -0.13
HealthMFIP Incentives Only
Parent report 4.16 4.23 -0.07 -0.06SSP
Parent report 4.10 4.01 0.10 ** 0.13
Full MFIPParent report 4.06 4.23 -0.17 -0.16
(continued)
101
Table 3 (continued)
Outcome
ProgramGroup
Average
ControlGroup
Average Impact
DifferenceBetween
Effect Size Subgroupof Impact Impacts
AchievementMFIP Incentives Only
Children aged 6-9
Parent report 3.80 3.52 0.28 0.23SSP
Parent report 3.68 3.63 0.05 0.05Math skills test 0.59 0.57 0.02 0.09
New HopeParent report 3.90 3.68 0.22 0.20Teacher report 3.17 3.03 0.14 0.14
Full MFIPParent report 3.88 3.52 0.36 ** 0.30
Externalizing behavior problemsMFIP Incentives Only
Parent report 5.18 7.50 -2.32 *** -0.38SSP
Parent report 1.25 1.25 0.00 -0.01New Hope
Parent report 2.42 2.41 0.01 0.02Teacher report 2.20 2.22 -0.01 -0.02
Full MFIPParent report 5.54 7.50 -1.96 ** -0.32
Positive behaviorMFIP Incentives Only
Parent report 194.80 186.03 8.77 0.20SSP
Parent report 2.60 2.58 0.01 0.03New Hope
Parent report 3.98 3.94 0.04 0.09Teacher report 3.58 3.47 0.11 0.16
Full MFIPParent report 195.17 186.03 9.14 0.21
HealthMFIP Incentives Only
Parent report 4.34 4.11 0.23 0.21SSP
Parent report 4.11 4.03 0.07 0.09Full MFIP
Parent report 3.98 4.11 -0.14 -0.13
(continued)
-7102
Table 3 (continued)
NOTES: In each study, children were selected for inclusion in the sample on the basis of their age at random assignmentor their age at follow-up.
The MFIP sample includes children (of parents in the MFIP evaluation) aged 6-12 at the time of the three yearfollow-up survey (aged approximately 3-9 at the time of random assignment) whose parents were long-term recipients inurban counties and underwent random assignment between April 1, 1994, and October 31, 1994 (sample size forestimates of Full MFIP = 488; sample size for MFIP Incentives Only = 472).
The SSP sample includes children (of parents in the SSP evaluation) aged 6-11 at the time of the three-yearsurvey (aged approximately 3-8 at random assignment) who were living in the home at the time of random assignmentand at the time of the three-year follow-up survey (sample size = 2,158).
The New Hope sample includes children of the single parents in the New Hope evaluation who were aged 3-9at random assignment and whose parents participated in the two-year follow-up survey (sample size = 546).
Because of rounding, the impacts shown may not exactly equal the differences between the program andcontrol group averages.
Statistical significance levels are indicated as: * = 10 percent; ** = 5 percent; *** = 1 percent (two-tailed test).A statistical test was performed to determine whether the differences between subgroup impacts were
statistically significant. The resulting statistical significance levels are indicated in the "Difference Between SubgroupImpacts" column as: * = 10 percent; ** = 5 percent; *** = 1 percent.
-78- 1 C 3
How Welfare and Work Policies Affect Children
Table 4
Earnings Supplement Programs
Impacts for Children Who Were Preschool-Aged or Early Elementary School-Agedat Random Assignment, by Children's Gender
Outcome
ProgramGroup
Average
ControlGroup
Average ImpactEffect Sizeof Impact
DifferenceBetween
SubgroupImpacts
AchievementMFIP Incentives Only
Boys
Parent report 3.83 3.77 0.06 0.05 ***
SSPParent report 3.55 3.50 0.05 0.05Math skills test 0.57 0.53 0.04 * 0.13
New HopeParent report 3.84 3.72 0.13 0.12Teacher report 3.27 2.92 0.35 ** 0.36
Full MFIPParent report 4.05 3.77 0.28 ** 0.24
Externalizing behavior problemsMFIP Incentives Only
Parent report 6.36 6.33 0.02 0.00 ***
SSPParent report 1.31 1.30 0.01 0.03
New HopeParent report 2.50 2.63 -0.13 -0.18Teacher report 2.09 2.52 -0.43 *** -0.49 ***
Full MFIPParent report 5.78 6.33 -0.56 -0.10
Positive behaviorMFIP Incentives Only
Parent report 193.83 194.59 -0.76 -0.02SSP
Parent report 2.52 2.52 -0.01 -0.02New Hope
Parent report 3.98 3.89 0.09 0.18 *
Teacher report 3.61 3.29 0.31 *** 0.46 **
Full MFIPParent report 191.35 194.59 -3.24 -0.09
HealthMFIP Incentives Only
Parent report 4.29 4.23 0.05 0.05
SSP
Parent report 4.05 4.01 0.04 0.05
Full MFIPParent report 4.12 4.23 -0.11 -0.10
(continued)
-79- 1 0 4
Table 4 (continued)
Outcome
ProgramGroup
Average
ControlGroup
Average ImpactEffect Size
of Impact
DifferenceBetween
SubgroupImpacts
AchievementMFIP Incentives Only
Girls
Parent report 4.42 4.15 0.27 ** 0.24 ***
SSPParent report 3.88 3.72 0.16 *** 0.17Teacher report 0.56 0.52 0.04 * 0.15
New HopeParent report 4.14 4.11 0.02 0.02Teacher report 3.39 3.27 0.12 0.12
Full MFIPParent report 4.20 4.15 0.05 0.04
Externalizing behavior problemsMFIP Incentives Only
Parent report 4.09 5.56 -1.47 ** -0.30 ***
SSPParent report 1.17 1.19 -0.02 -0.07
New HopeParent report 2.62 2.54 0.08 0.11Teacher report 2.15 1.89 0.26 ** 0.30 ***
Full MFIPParent report 4.49 5.56 -1.07 * -0.22
Positive behaviorMFIP Incentives Only
Parent report 207.33 191.57 15.76 *** 0.38 *
SSPParent report 2.66 2.65 0.00 0.00
New HopeParent report 3.94 4.01 -0.07 -0.15 *
Teacher report 3.71 3.72 0.00 -0.01 **
Full MFIPParent report 198.59 191.57 7.02 0.17
HealthMFIP Incentives Only
Parent report 4.26 4.16 0.09 0.09SSP
Parent report 4.16 4.02 0.14 *** 0.17Full MFIP
Parent report 4.11 4.16 -0.06 -0.05
-80- 1 G 5
(continued)
Table 4 (continued)
NOTES: In each study, children were selected for inclusion in the sample on the basis of their age at random
assignment or their age at follow-up.The MFIP sample includes children (of parents in the MFIP evaluation) aged 5-12 at the time of the three-year
follow-up survey (aged approximately 2-9 at the time of random assignment) whose parents were long-term recipientsin urban counties and underwent random assignment between April 1, 1994, and October 31, 1994 (sample size forestimates of Full MFIP = 587; sample size for MFIP Incentives Only = 573).
The SSP sample includes children (of parents in the SSP evaluation) aged 6-11 at the time of the three-yearsurvey (aged approximately 3-8 at random assignment) who were living in the home at the time of random assignment
and at the time of the three-year follow-up survey (sample size = 2,158).The New Hope sample includes children of the single parents in the New Hope evaluation who were aged 1-
10 at random assignment and whose parents participated in the two-year follow-up survey (sample size = 832).Because of rounding, the impacts shown may not exactly equal the differences between the program and
control group averages.Statistical significance levels are indicated as: * = 10 percent; ** = 5 percent; *** = 1 percent (two-tailed test).A statistical test was performed to determine whether the differences between subgroup impacts were
statistically significant. The resulting statistical significance levels are indicated in the "Difference Between SubgroupImpacts" column as: * = 10 percent; ** = 5 percent; *** = 1 percent.
10:6-81-
How Welfare and Work Policies Affect Children
Table 5
Programs with Mandatory Employment Services
Impacts for Children Who Were Preschool-Aged at Random Assignment
Outcome
ProgramGroup
Average
ControlGroup
Average ImpactEffect Size
of Impact
AchievementEducation-first programs
Altanta 7.66 7.36 0.30 0.10Grand Rapids 7.48 7.40 0.08 0.03Riverside 7.23 7.10 0.13 0.03
Job-search-first programsAltanta 7.92 7.35 0.58 *** 0.19Grand Rapids 7.40 7.30 0.09 0.03Riverside 7.36 7.52 -0.15 -0.04
Externalizing behavior problemsEducation-first programs
Altanta 0.42 0.46 -0.04 * -0.10Grand Rapids 0.50 0.44 0.05 0.15Riverside 0.43 0.41 0.02 0.05
Job-search-first programsAltanta 0.40 0.47 -0.06 *** -0.15Grand Rapids 0.50 0.44 0.06 * 0.17Riverside 0.41 0.40 0.01 0.02
Positive behaviorEducation-first programs
Altanta 1.50 1.50 0.00 0.01Grand Rapids 1.62 1.60 0.02 0.05Riverside 1.56 1.59 -0.02 -0.05
Job-search-first programsAltanta 1.51 1.50 0.01 0.03Grand Rapids 1.61 1.60 0.01 0.04Riverside 1.63 1.62 0.02 0.03
HealthEducation-first programs
Altanta 4.25 4.27 -0.02 -0.02Grand Rapids 4.35 4.26 0.09 0.09Riverside 3.96 4.16 -0.20 ** -0.17
Job-search-first programsAltanta 4.25 4.27 -0.02 -0.02Grand Rapids 4.19 4.26 -0.07 -0.07Riverside 3.97 4.23 -0.26 *** -0.21
(continued)
-82- 4 ^
Table 5 (continued)
NOTES: The NEWWS sample includes children of single mothers in the NEWWS evaluation aged 3-
5 at the beginning of the study whose parents were randomly selected to participate in the two-year
follow-up survey (sample sizes for education-first programs: Atlanta = 1,026, Grand Rapids = 421,
Riverside = 578; sample sizes for job-search-first programs: Atlanta = 902, Grand Rapids = 441,
Riverside = 694).Because of rounding, the impacts shown may not exactly equal the differences between the
program and control group averages.Statistical significance levels are indicated as: * = 10 percent; ** = 5 percent; *** = 1
percent (two-tailed test).
-83-
How Welfare and Work Policies Affect Children
Table 6
Programs with Mandatory Employment Services
Impacts for Children of Long-Term Recipients Who Were Preschool-Agedat Random Assignment
Outcome
ProgramGroup
Average
AchievementEducation-first programs
Altanta 7.41Grand Rapids 7.06Riverside 7.28
Job-search-first programsAltanta 7.82Grand Rapids 7.08Riverside 6.97
Externalizing behavior problems
Education-first programsAltanta 0.43Grand Rapids 0.50Riverside 0.45
Job-search-first programsAltanta 0.41Grand Rapids 0.50Riverside 0.45
Positive behaviorEducation-first programs
Altanta 1.48Grand Rapids 1.57Riverside 1.58
Job-search-first programsAltanta 1.47Grand Rapids 1.62Riverside 1.58
HealthEducation-first programs
Altanta 4.21Grand Rapids 4.30Riverside 3.91
Job-search-first programsAltanta 4.26Grand Rapids 4.18Riverside 3.89
-84-
ControlGroup
Average ImpactEffect Size
of Impact
6.86 0.55 *** 0.207.19 -0.14 -0.057.23 0.05 0.01
6.86 0.96 *** 0.347.14 -0.06 -0.027.60 -0.63 ** -0.15
0.50 -0.07 ** -0.160.45 0.05 0.140.40 0.06 0.14
0.50 -0.10 *** -0.230.44 0.05 0.150.38 0.07 ** 0.18
1.45 0.02 0.051.60 -0.03 -0.091.59 -0.02 -0.04
1.45 0.02 0.041.60 0.03 0.071.61 -0.03 -0.06
4.21 0.00 0.004.20 0.10 0.104.20 -0.29 *** -0.26
4.21 0.05 0.054.23 -0.05 -0.054.24 -0.35 *** -0.30
(continued)
109
Table 6 (continued)
NOTES: The NEWWS sample includes children of single mothers inthe NEWWS evaluation aged 3-
5 at the beginning of the study whose parents had at least two years of welfare receipt prior to random
assignment and were randomly selected to participate in the two-year follow-up survey (sample sizes
for education-first programs: Atlanta = 757, Grand Rapids = 296, Riverside = 425; sample sizes for job.
search-first programs: Atlanta = 669, Grand Rapids = 327, Riverside = 459).
Because of rounding, the impacts shown may not exactly equal the differences between the
program and control group averages.Statistical significance levels are indicated as: * = 10 percent; ** = 5 percent; *** = 1
percent (two-tailed test).
110-85-
How Welfare and Work Policies Affect Children
Table 7
Time-Limited Program (FTP)
Impacts for Children Who Were Preschool-Aged or Elementary School-Agedat Random Assignment
Outcome
ProgramGroup
Average
ControlGroup
Average ImpactEffect Size
of Impact
Achievement 4.09 3.98 0.10 0.09
Externalizing behavior problems 4.33 4.28 0.06 0.01
Positive behavior 59.04 60.22 -1.17 * -0.11
Health 4.23 4.14 0.09 * 0.09
NOTES: The FTP sample includes children of single parents in the FTP evaluation aged 5-12 at thetime of the four-year follow-up survey (aged approximately 1-8 at the time of random assignment)whose parents underwent random assignment between August 1994 and February 1995 andparticipated in the four-year follow-up survey (sample size = 1,108).
Because of rounding, the impacts shown may not exactly equal the differences between theprogram and control group averages.
Statistical significance levels are indicated as: * = 10 percent; **= 5 percent; *** = 1percent (two-tailed test).
- 8 6 -
How Welfare and Work Policies Affect Children
Table 8
Two Different Programs (SSP and FTP)
Impacts for Adolescent Children
Outcome
ProgramGroup
Average
ControlGroup
Average ImpactEffect Sizeof Impact
SSPAchievement
Parent report 3.43 3.54 -0.11 * -0.11
Adolescent report 3.50 3.57 -0.07 -0.09
School behavior 1.40 1.34 0.06 * 0.09
Smoking 26.52 22.13 4.39 * 0.11
Drinking 8.91 4.65 4.27 *** 0.20
Health 4.10 4.13 -0.04 -0.05
FTP
Achievement 3.70 3.90 -0.20 * -0.14
Ever suspended 40.70 32.70 8.00 ** 0.17
Ever arrested 9.60 9.20 0.40 0.01
Ever had a baby 2.80 3.30 -0.50 -0.03
NOTES: The SSP sample includes children of single parents in the SSP evaluation aged 12-18 at thetime of the three-year follow-up survey (aged approximately 9-15 at random assignment) who wereliving in the home at the time of random assignment and at the time of the three-year follow-upsurvey (sample size = 1,417).
The FTP sample includes children of single parents in the FTP evaluation aged 13-17 at thetime of the four-year follow-up survey (aged approximately 9-13 at random assignment) whoseparents underwent random assignment between August 1994 and February 1995 and participated inthe four-year follow-up survey (sample size = 741).
Because of rounding, the impacts shown may not exactly equal the differences between theprogram and control group averages.
Statistical significance levels are indicated as: * = 10 percent; ** = 5 percent; *** = 1percent (two-tailed test).
112-87-
References
Alvarez, W. 1985. "The Meaning of Maternal Employment for Mothers and the Perceptions of TheirThree-Year-Old Children." Child Development, 56: 350-360.
Berlin, G. L. 2000. Encouraging Work, Reducing Poverty: The Impact of Work Incentive Programs. NewYork: Manpower Demonstration Research Corporation.
Bloom, D., J. J. Kemple, P. Morris, S. Scrivener, N. Verma, and R. Hendra. 2000. The Family TransitionProgram: Final Report on Florida's Initial Time-Limited Welfare Program. New York: Man-power Demonstration Research Corporation.
Bloom, D., L. Melton, C. Michalopoulos, S. Scrivener, and J. Walter. 2000. Implementation and EarlyImpacts of Connecticut's Welfare Reform Initiative. New York: Manpower Demonstration Re-search Corporation.
Bloom, D., and C. Michalopoulos. 2001. How Welfare and Work Policies Affect Employment and In-come: A Synthesis of Research. New York: Manpower Demonstration Research Corporation.
Bolger, K. E., C. J. Patterson, W. W. Thompson, and J. B. Kupersmidt. 1995. "Psychosocial AdjustmentAmong Children Experiencing Persistent and Intermittent Family Economic Hardship." Child De-velopment, 66: 1107-1129.
Bos, J., A. Huston, R. Granger, G. Duncan, T. Brock, and V. McLoyd. 1999. New Hope for People withLow Incomes. New York: Manpower Demonstration Research Corporation.
Bradley, R. H., and B. M. Caldwell. 1984. "174 Children: A Study of the Relation Between the HomeEnvironment and Early Cognitive Development in the First 5 Years." In The Home Environmentand Early Cognitive Development, ed. A. Gottfried, pp. 5-56. Orlando, FL: Academic Press.
Burchinal, M. R., J. E. Roberts, R. Riggins, Jr., S. A. Zeisel, E. Neebe, and D. Bryant. 2000. "RelatingQuality of Center-Based Child Care to Early Cognitive and Language Development Longitudi-nally." Child Development, 71: 339-357.
Butler, A. 1990. "The Effects of Welfare Guarantees on Children's Educational Attainment." Social Sci-ence Research, 19: 175-203.
Caspi, A., B. R. Wright, T. E. Moffit, and P. A. Silva. 1998. "Early Failure in the Labor Market: Child-hood and Adolescent Predictors of Unemployment in the Transition to Adulthood." American So-ciological Review, 63: 424-451.
Crouter, A. C., M. F. Bumpus, and S. M. McHale. 1999. "Linking Parents' Work Pressure and Adoles-cents' Well-Being: Insights into Dynamics in Dual-Earner Families." Developmental Psychology,35: 1453-1461.
Duncan, G. J., and J. Brooks-Gunn, eds. 1997. Consequences of Growing Up Poor. New York: RussellSage Foundation.
Duncan, G. J., J. Brooks-Gunn, and P. K. Klebanov. 1994. "Economic Deprivation and Early ChildhoodDevelopment." Child Development, 65: 296-318.
113-89-
Duncan, G. J., and W. Yeung. 1995. "Extent and Consequences of Welfare Dependence Among Amer-ica's Children." Children and Youth Services Review, 17(1/2): 1-26.
Entwisle, D. R. 1985. "The Role of Schools in Sustaining Early Childhood Program Benefits." Future ofChildren, 5(3): 133-144
Farel, A. M. 1980. "Effects of Preferred Maternal Roles, Maternal Employment, and SociodemographicStatus on School Adjustment and Competence." Child Development, 51: 1179-1196.
Freedman, S., D. Friedlander, G. Hamilton, J. Rock, M. Mitchell, J. Nudelman, A. Schweder, and L.Storto. 2000. Evaluating Alternative Welfare-to-Work Approaches: Two-Year Impacts for ElevenPrograms. National Evaluation of Welfare-to-Work Strategies. Washington, DC: U.S. Departmentof Health and Human Services, Administration for Children and Families and Office of the Assis-tant Secretary for Planning and Evaluation; U.S. Department of Education, Office of the UnderSecretary and Office of Vocational and Adult Education.
Gennetian, L., and C. Miller. 2000. Reforming Welfare and Rewarding Work: Final Report of the Minne-sota Family Investment Program, Vol. 2, Effects on Children. New York: Manpower Demonstra-tion Research Corporation.
Golombok, S., and R. Fivush. 1994. Gender Development. New York: Cambridge University Press.
Gomby, D. S., P. L. Culross, and R. E. Behrman. 1999. "Home Visiting: Recent Program EvaluationsAnalysis and Recommendations." Future of Children, 9(1): 4-26.
Hamilton, G., with S. Freedman and S. M. McGroder. 2000. Do Mandatory Welfare-to-Work ProgramsAffect the Well-Being of Children? A Synthesis of Child Research Conducted as Part of the Na-tional Evaluation of Welfare-to-Work Strategies. Washington, DC: U.S Department of Health andHuman Services, Administration for Children and Families and Office of the Assistant Secretaryfor Planning and Evaluation; and U.S. Department of Education, Office of the Under Secretary andOffice of Vocational and Adult Education.
Hamilton, G., T. Brock, M. Farrell, D. Friedlander, and K. Harknett. 1997. Evaluating Two Welfare-to-Work Program Approaches: Two-Year Findings on the Labor Force Attachment and Human Capi-tal Development Programs in Three Sites. National Evaluation of Welfare-to-Work Strategies.Washington, DC: U.S. Department of Health and Human Services, Administration for Childrenand Families and Office of the Assistant Secretary for Planning and Evaluation; and U.S. Depart-ment of Education, Office of the Under Secretary and Office of Vocational and Adult Education.
Harvey, E. 1999. "Short-Term and Long-Term Effects of Early Parental Employment on Children of theNational Longitudinal Survey of Youth. "Developmental Psychology, 35(2): 445-459.
Haskins, R. Forthcoming, 2001. "Liberal and Conservative Influences on the Welfare Reform Legislationof 1996." In G. J. Duncan and L. Chase-Lansdale, For Better and for Worse: Welfare Reform andthe Well-Being of Children and Families. New York: Russell Sage Foundation.
Haveman, R., and B. Wolfe. 1995. "The Determinants of Children's Attainments." Journal of EconomicLiterature, 33(4): 1829-1878.
Hofferth, S. L., J. Smith, V. C. McLoyd, and J. Finkelstein. 2000. Achievement and Behavior AmongChildren of Welfare Recipients, Welfare Leavers, and Low Income Single Mothers. Ann Arbor:University of Michigan, Institute for Social Research.
"-90- 114
Huston, A. C. 1983. "Sex Typing." In Handbook of Child Psychology, 4th ed., Vol. 4, Socialization, Per-sonality, and Social Development, ed. P. Mussen (Series) and E. M. Hetherington (Volume), pp.387-467. New York: Wiley.
Huston, A., G. Duncan, R. Granger, H. Bos, V. McLoyd, R. Mistry, D. Crosby, C. Gibson, K. Magnus-son, J. Romich, and A. Ventura. Forthcoming, 2001. "Work-Based Anti-Poverty Programs for Par-ents can Enhance the School Performance and Social Behavior of Children." Child Development.
Infant Health and Development Program (IHDP). 1990. "Enhancing the Outcomes of Low-Birth-Weight,Premature Infants." Journal of the American Medical Association, 263(22): 3035-3042.
Jackson, A. P. 1993. "Black, Single, Working Mothers in Poverty: Preferences for Employment, Well-Being, and Perceptions of Children." Social Work, 38: 26-34.
Kaestner, R., and H. Corman. 1995. The Impact of Child Health and Family Inputs on Child CognitiveDevelopment. NBER. Working Paper 5257. Cambridge, MA: National Bureau of Economic Re-search.
Klerman, L. 1991. "The Health of Poor Children: Problems and Programs." In Children and Poverty, ed.A. Huston, Cambridge, Eng., and New York: Cambridge University Press.
Knox, V., C. Miller, and L. Gennetian. 2000. Reforming Welfare and Rewarding Work: A Summary ofthe Final Report on the Minnesota Family Investment Program. New York: Manpower Demonstra-tion Research Corporation.
Korenman, S., and J. E. Miller. 1997. "Effects. of Long-Term Poverty on Physcial Health of Children inthe National Longitudinal Survey of Youth." In Consequences of Growing Up Poor, ed. G. J. Dun-can and J. Brooks-Gunn, pp. 70-99. New York: Russell Sage Foundation.
Krueger, A. 1997. "Experimental Estimates of Education Production Functions." Working Paper. Prince-ton, NJ: Princeton University.
Laosa, L. 1983. "School, Occupation, Culture, and Family." In Changing Families, ed. E. Sigel and L.Laosa, pp.79-135. New York: Plenum Press.
Lanier, M. B., L. Zippiroli, and R. E. Behrman. 1999. "When School Is Out: Analysis and Recommenda-tions." Future of Children, 9(2): 4-20.
Levine, P. B., and D. J. Zimmerman. 2000. "Children's Welfare Exposure and Subsequent Develop-ment." JCPR Working Paper 130. Evanston, IL, and Chicago: Northwestern University and Uni-versity of Chicago, Joint Center for Poverty Research.
Marshall, N. L., C. Garcia, F. Marx, K. McCartney, N. Keefe, and J. Ruh. 1997. "After-School Time andChildren's Behavioral Adjustment." Merrill-Palmer Quarterly, 43(3): 497-514.
Masten, A. S., J. D. Coatsworth, J. Neemann, S. D. Gest, A. Tellegen, and N. Garmezy. 1995. "TheStructure and Coherence of Competence from Childhood Through Adolescence." Child Develop-ment, 66: 1635-1659.
Mayer, S. E. 1997. What Money Can't Buy: Family Income and Children's Life Chances. Cambridge,MA: Harvard University Press.
McGroder, S. M., M. J. Zaslow, K. A. Moore, and S. M. LeMenestrel. 2000. Impacts on YoungChildren and Their Families Two Years After Enrollment: Findings from the Child Out-comes Study. National Evaluation of Welfare-to-Work Strategies. Washington, DC: U.S.Department of Health and Human Services, Office of the Assistant Secretary for Planningand Evaluation and Administration for Children and Families; and U.S. Department ofEducation, Office of the Under Secretary and Office of Vocational and Adult Education.
McLoyd, V. 1998. "Children in Poverty: Development, Public Policy, and Practice." In Handbook ofChild Psychology, 4th ed., ed. I. E. Siegel and K. A. Renninger. New York: Wiley.
McLoyd, V. C., T. E. Jayartne, R. Ceballo, and J. Borquez. 1994. "Unemployment and Work InterruptionAmong African-American Single Mothers: Effects on Parenting and Adolescent Socio-EmotionalFunctioning." Child Development, 65: 562-589.
Michalopoulos, C., D. Card, L. Gennetian, K. Harknett, and P. K. Robins. 2000. The Self-SufficiencyProject at 36 Months: Effects of a Financial Work Incentive on Employment and Income. Ottawa:Social Research and Demonstration Corporation.
Miller, C., V. Knox, L. Gennetian, M. Dodoo, J. Hunter, and C. Redcross. 2000. Reforming Welfare andRewarding Work: Final Report on the Minnesota Family Investment Program, Vol. 1, Effects onAdults. New York: Manpower Demonstration Research Corporation.
Moore, K., and A. Driscoll. 1997. "Low-Wage Maternal Employment and Outcomes for Children: AStudy." Future of Children, 7(1): 122-127.
Moore, K. A., D. R. Morrison, M. J. Zaslow, and D. A. Glei. 1994. "Ebbing and Flowing, Learning andGrowing: Family Economic Resources and Children's Development." Paper prepared for a work-shop on Welfare and Child Development, National Academy of Sciences, Washington, DC.
Morris, P., and C. Michalopoulos. 2000. The Self-Sufficiency Project at 36 Months: Effects on Childrenof a Program That Increased Parental Employment and Income. Ottawa: Social Research andDemonstration Corporation.
NICHD Early Child Care Research Network. 1998a. "Relations Between Family Predictors and ChildOutcomes: Are They Weaker for Children in Child Care?" Developmental Psychology, 34: 1119-1128.
NICHD Early Child Care Research Network. 1998b. "Early Child Care and Compliance, Cooperation,Defiance and Problem Behavior at 24 and 36 Months." Child Development, 46: 1145-1170.
NICHD Early Child Care Research Network. 2000. "The Relation of Child Care to Cognitive and Lan-guage Development." Child Development, 71: 960-980.
NICHD Early Child Care Research Network. Forthcoming, 2001. "Child Care and Common Communi-cable Illnesses: Results from the NICHD Study of Early Child Care." Archives of Pediatrics andAdolescent Medicine.
Parcel, T. L., and E. G. Menaghan, 1994. Parents' Jobs and Children 's Lives. New York: Aldine deGruyter.
Parcel, T. L., and E. G. Menaghan. 1997. "Effects of Low-Wage Employment on Family Well-Being."Future of Children, 7(1): 116-121.
116
Peisner-Feinberg, E. S., M. R. Burchinal, R. M. Clifford, M. L. Culkin, C. Howes, S. L. Kagan, N. Yazs-jian, P. By ler, J. Rustici, and J. Zelazo. 1999. "The Children of the Cost, Quality, and OutcomesStudy Go to School." Technical Report. Chapel Hill: University of North Carolina.
Pettit, G. S., J. E. Bates, K. A. Dodge, and D. W. Meece. 1999. "The. Impact of After-School Peer Con-tact on Early Adolescent Externalizing Problems Is Moderated by Parental Monitoring, PerceivedNeighborhood Safety and Prior Adjustment." Child Development, 70(3): 768-778.
Pettit, G. S., Laird, R. D., Bates, J. D., and Dodge, K. A. (1997). "Patterns of After-School Care in MiddleChildhood: Risk Factors and Developmental Outcomes." Merrill Palmer Quarterly, 43: 515-538.
Ramey, C. T., F. A. Campbell, M. R. Burchinal, M. L. Skinner, D. M. Gardner, and S. L. Ramey. 2000."Persistent Effects of Early Childhood Education on High-Risk Children and Their Mothers." Ap-plied Developmental Science, 4: 2-14.
Ratcliffe, C. 1996. "Intergenerational Transmission of Welfare Participation: How Large Is the CausalLink?" Mimeo. Ithaca, NY: Cornell University, Department of Economics.
Rosenweig, M. R., and K. I. Wolpin. 1994. "Are There Increasing Returns to the Intergenerational Pro-duction of Human Capital? Maternal Schooling and Child Intellectual Development." Journal ofHuman Resources, 29: 670-693.
Sampson, R. J., and J. H. Laub. 1994. "Urban Poverty and Family Context of Delinquency: A New Lookat Structure and Process in a Classical Study." Child Development, 65: 523-540.
Shonkoff, J. P., and D. A. Phillips. 2000. From Neurons to Neighborhoods: The Science of Early Child-hood Development. Washington, DC: National Academy Press.
Smith, J. R., and J. Brooks-Gunn. 1994. "Developmental Effects of Natural Transitions in Welfare Re-ceipt Paper prepared for a workshop on Welfare and Child Development, National Academy ofSciences, Washington, DC.
Smith, J. R., J. Brooks-Gunn, and P. K. Klebanov. 1997. "Consequences of Living in Poverty for YoungChildren's Cognitive and Verbal Ability and Early School Achievement." In Consequences ofGrowing Up Poor, ed. G. J. Duncan and J. Brooks-Gunn. New York: Russell Sage Foundation.
Sugland, B. W., M. J. Zaslow, J. R. Smith, J. Brooks-Gunn, K. A. Moore, C. Blumenthal, T. Griffin, andR. Bradley. 1995. "The Early Childhood HOME Inventory and HOME Short Form in DifferingSocio-Cultural Groups: Are There Differences in Underlying Structure, Internal Consistency ofSubscales, and Patterns of Prediction?" Journal of Family Issues 16(5): 632-663.
Vandell, D. L., and J. Ramanan. 1992. "Effects of Early and Recent Maternal Employment on Childrenfrom Low-Income Families." Child Development, 63(4): 938-949.
Word, E., J. Johnston, H. P. Bain, B. D. Fulton, J. B. Zaharias, C. M. Achilles, M. N. Lintz, J. Folger, andC. Breda. 1990. The State of Tennessee's Student/Teacher Achievement Ratio (STAR) Project FinalReport. Nashville: Tennessee Department of Education.
Yoshikawa, H. 1995. "Long-Term Effects of Early Childhood Programs on Social Outcomes and Delin-quency." Future of Children, 5(3): 51-75.
Yoshikawa, H. 1999. "Welfare Dynamics, Support Services, Mothers' Earnings, and Child CognitiveDevelopment: Implications for Contemporary Welfare Reform." Child Development, 70: 779-801.
dt P-p
93
Zaslow, M. J., and C. A. Emig. 1997. "When Low-Income Mothers Go to Work: Implications for Chil-dren." Future of Children, 7(1): 110-115.
Zaslow, M. J., S. McGroder, G. Cave, and C. Mariner. 1999. "Maternal Employment and Measures ofChildren's Health and Development Among Families with Some History of Welfare Receipt." Re-search in the Sociology of Work, 7: 233-259.
Zaslow, M. J., S. McGroder, K. A. Moore, and S. LeMenestrel. 1999. "Behavior Problems and CognitiveSchool Readiness Among Young Children in Families with a History of Welfare Receipt: Diverg-ing Patterns of Development and Their Predictors." Paper presented at the meetings of the Societyof Research in Child Development, Albuquerque, New Mexico, April.
Zaslow, M. J., E. Oldham, K. A. Moore, and E. Magenheim. 1998. "Welfare Families' Use of EarlyChildhood Care and Education Programs and Implications for Children." Early Childhood Re-search Quarterly, 13: 535-563.
Zill, N., K. A. Moore, E. Wolpow Smith, T. Stief, and M. J. Coiro. 1995. "The Life Circumstances andDevelopment of Children in Welfare Families: A Profile Based on National Survey Data." In Es-cape from Poverty, ed. P. Lindsay Chase-Lansdale and Jeanne Brooks-Dunn. Cambridge, Eng.:Cambridge University Press.
1 8:94-
Recent Publications on MDRC Projects
Note: For works not published by MDRC, the publisher's name is shown in parentheses. A complete publications listis available from MDRC and on its Web site (www.mdrc.org); which also contains copies of MDRC'spublications.
Reforming Welfare and MakingWork Pay
Next Generation ProjectA collaboration among researchers at MDRC andseveral leading research institutions focused on studyingthe effects of welfare, antipoverty, and employmentpolicies on children and families.
How Welfare and Work Policies Affect Children: ASynthesis of Research. 2001. Pamela Morris, AlethaHuston, Greg Duncan, Danielle Crosby, JohannesBos.
How Welfare and Work Policies Affect Employment andIncome: A Synthesis of Research. 2001. Dan Bloom,Charles Michalopoulos.
ReWORKing Welfare: Technical Assistance forStates and LocalitiesA multifaceted effort to assist states and localities indesigning and implementing their welfare reformprograms. The project includes a series of "how-to"guides, conferences, briefings, and customized, in-depthtechnical assistance.
After AFDC: Welfare-to-Work Choices and Challengesfor States. 1997. Dan Bloom.
Changing to a Work First Strategy: Lessons from LosAngeles County's GAIN Program for WelfareRecipients. 1997. Evan Weissman.
Work First: How to Implement an Employment-FocusedApproach to Welfare Reform. 1997. Amy Brown.
Business Partnerships: How to Involve Employers inWelfare Reform. 1998. Amy Brown, Maria Buck, ErikSkinner.
Learnfare: How to Implement a Mandatory Stay-in-School Program for Teenage Parents on Welfare.1998. David Long, Johannes Bos.
Promoting Participation: How to Increase Involvementin Welfare-to-Work Activities. 1999. Gayle Hamilton,Susan Scrivener.
Encouraging Work, Reducing Poverty: The Impact ofWork Incentive Programs. 2000. Gordon Berlin.
Steady Work and Better Jobs: How to Help Low-IncomeParents Sustain Employment and Advance in theWorkforce. 2000. Julie Strawn, Karin Martinson.
Project on Devolution and Urban ChangeA multi-year study in four major urban countiesCuyahoga County, Ohio (which includes the city ofCleveland), Los Angeles, Miami-Dade, and Philadelphia
that examines how welfare reforms are being
implemented and affect poor people, theirneighborhoods, and the institutions that serve them.
Big Cities and Welfare Reform: Early Implementationand Ethnographic Findings from the Project onDevolution and Urban Change. 1999. Janet Quint,Kathryn Edin, Maria Buck, Barbara Fink, YolandaPadilla, Olis Simmons-Hewitt, Mary Valmont.
Food Security and Hunger in Poor, Mother-HeadedFamilies in Four U.S. Cities. 2000. Denise Polit,Andrew London, John Martinez.
Assessing the Impact of Welfare Reform on UrbanCommunities: The Urban Change Project andMethodological Considerations. Forthcoming.Charles Michalopoulos, Johannes Bos, RobertLalonde, Nandita Verma.
Post-TANF Food Stamp and Medicaid Benefits: FactorsThat Aid or Impede Their Receipt. 2001. Janet Quint,Rebecca Widom.
Time Limits
Florida's Family Transition ProgramAn evaluation of Florida's initial time-limited welfareprogram, which includes services, requirements, andfinancial work incentives intended to reduce long-termwelfare receipt and help welfare recipients fmd.andkeep jobs.
The Family Transition Program: An EarlyImplementation Report on Florida's Time-LimitedWelfare Initiative. 1995. Dan Bloom.
The Family Transition Program: Implementation andEarly Impacts of Florida's Initial Time-LimitedWelfare Program. 1997. Dan Bloom, James Kemple,Robin Rogers-Dillon.
The Family Transition Program: Implementation andInterim Impacts of Florida's Initial Time-LimitedWelfare Program. 1998. Dan Bloom, Mary Farrell,James Kemple, Nandita Verma.
The Family Transition Program: Implementation andThree-Year Impacts of Florida's Initial Time-LimitedWelfare Program. 1999. Dan Bloom, Mary Farrell,James Kemple, Nandita Verma.
Cross-State Study of Time-Limited WelfareAn examination of the implementation of some of thefirst state-initiated time-limited welfare programs.
Implementing Time-Limited Welfare: Early Experiencesin Three States. 1995. Dan Bloom, David Butler.
-95-11 9
The View from the Field: As Time Limits Approach,Welfare Recipients and Staff Talk About TheirAttitudes and Expectations. 1997. Amy Brown, DanBloom, David Butler.
Welfare Time Limits: An Interim Report Card. 1999.Dan Bloom.
Connecticut's Jobs First ProgramAn evaluation of Connecticut's statewide time-limitedwelfare program, which includes financial workincentives and requirements to participate inemployment-related services aimed at rapid jobplacement. This study provides some of the earliestinformation on the effects of time limits in major urbanareas.
Early Data on the Implementation of Connecticut's JobsFirst Program. 1997. Dan Bloom, Mary Andes.
Jobs First: Early Implementation of Connecticut'sWelfare Reform Initiative. 1998. Dan Bloom, MaryAndes, Claudia Nicholson.
Connecticut Post-Time Limit Tracking Study: Three-Month Survey Results. 1998. Jo Anna Hunter-Maims,Dan Bloom, Richard Hendra, Johanna Walter.
Connecticut Post-Time Limit Tracking Study: Six-MonthSurvey Results. 1999. Jo Anna Hunter-Maims, DanBloom.
Jobs First: Implementation and Early Impacts ofConnecticut's Welfare Reform Initiative. 2000. DanBloom, Laura Melton, Charles Michalopoulos, SusanScrivener, Johanna Walter.
Vermont's Welfare Restructuring ProjectAn evaluation of Vermont's statewide welfare reformprogram, which includes a work requirement after acertain period of welfare receipt, and financial workincentives.
WRP: IMplementation and Early Impacts of Vermont'sWelfare Restructuring Project. 1998. Dan Bloom,Charles Michalopoulos, Johanna Walter, PatriciaAuspos.
Forty-Two Month Impacts of Vermont's WelfareRestructuring Project. 1999. Richard Hendra, CharlesMichalopoulos.
WRP: Key Findings from the Forty-Two-Month ClientSurvey. 2000. Dan Bloom, Richard Hendra, CharlesMichalopoulos.
Financial IncentivesEncouraging Work, Reducing Poverty: The Impact of
Work Incentive Programs. 2000. Gordon Berlin.
Minnesota Family Investment ProgramAn evaluation of Minnesota's pilot welfare reforminitiative, which aims to encourage work, alleviatepoverty, and reduce welfare dependence.
MFIP: An Early Report on Minnesota's Approach toWelfare Reform. 1995. Virginia Knox, Amy Brown,Winston Lin.
Making Welfare Work and Work Pay: Implementationand 18 -Month Impacts of the Minnesota Family
Investment Program. 1997. Cynthia Miller, VirginiaKnox, Patricia Auspos, Jo Anna Hunter-Manns, AlanOrenstein.
Reforming Welfare and Rewarding Work: Final Reporton the Minnesota Family Investment Program. 2000:
Volume 1: Effects on Adults. Cynthia Miller,Virginia Knox, Lisa Germetian, Martey Dodoo, JoAnna Hunter, Cindy Redcross.Volume 2: Effects on Children. Lisa Gennetian,Cynthia Miller.
Reforming Welfare and Rewarding Work: A Summaryof the Final Report on the Minnesota FamilyInvestment Program. 2000. Virginia Knox, CynthiaMiller, Lisa Gennetian.
Final Report on the Implementation and Impacts of theMinnesota Family Investment Program in RamseyCounty. 2000. Patricia Auspos, Cynthia Miller, JoAnna Hunter.
New Hope ProjectA test of a community-based, work-focused antipovertyprogram and welfare alternative operating in Milwaukee.
The New Hope Offer: Participants in the New HopeDemonstration Discuss Work, Family, and Self-Sufficiency. 1996. Dudley Benoit.
Creating New Hope: Implementation of a Program toReduce Poverty and Reform Welfare. 1997. ThomasBrock, Fred Doolittle, Veronica Fellerath, MichaelWiseman.
Who Got New Hope? 1997. Michael Wiseman.An Early Look at Community Service Jobs in the New
Hope Demonstration. 1998. Susan Poglinco, JulianBrash, Robert Granger.
New Hope for People with Low Incomes: Two-YearResults of a Program to Reduce Poverty and ReformWelfare. 1999. Johannes Bos, Aletha Huston, RobertGranger, Greg Duncan, Thomas Brock, VonnieMcLoyd.
Canada's Self-Sufficiency ProjectA test of the effectiveness of a temporary earningssupplement on the employment and welfare receipt ofpublic assistance recipients. Reports on the Self-Sufficiency Project are available from: Social Researchand Demonstration Corporation (SRDC), 275 Slater St.,Suite 900, Ottawa, Ontario KW 5H9, Canada. Tel.:613-237-4311; Fax: 613-237-5045. In the United States,the reports are also available from MDRC.
Creating an Alternative to Welfare: First-Year Findingson the Implementation, Welfare Impacts, and Costs ofthe Self-Sufficiency Project (Social Research andDemonstration Corporation [SRDC]). 1995. TodMijanovich, David Long.
The Struggle for Self-Sufficiency: Participants in theSelf-Sufficiency Project Talk About Work, Welfare,and Their Futures (SRDC). 1995. Wendy Bancroft,Sheila Currie Vernon.
12. -96-
Do Financial Incentives Encourage Welfare Recipientsto Work? Initial 18 -Month Findings from the Self-Sufficiency Project (SRDC). 1996. David Card, PhilipRobins.
When Work Pays Better Than Welfare: A Summary ofthe Self-Sufficiency Project's Implementation, FocusGroup, and Initial 18-Month Impact Reports (SRDC).1996.
How Important Are "Entry Effects" in FinancialIncentive Programs for Welfare Recipients?Experimental Evidence from the Self-SufficiencyProject (SRDC). 1997. David Card, Philip Robins,Winston Lin.
Do Work Incentives Have Unintended Consequences?Measuring "Entry Effects" in the Self-SufficiencyProject (SRDC). 1998. Gordon Berlin, WendyBancroft, David Card, Winston Lin, Philip Robins.
When Financial Incentives Encourage Work: Complete18 -Month Findings from the Self-Sufficiency Project(SRDC). 1998. Winston Lin, Philip Robins, DavidCard, Kristen Harknett, Susanna Lui-Gurr.
Does SSP Plus Increase Employment? The Effect ofAdding Services to the Self-Sufficiency Project'sFinancial Incentives (SRDC). 1999. Gail Quets,Philip Robins, Elsie Pan, Charles Michalopoulos,David Card.
When Financial Work Incentives Pay for Themselves:Early Findings from the Self-Sufficiency Project'sApplicant Study (SRDC). 1999. CharlesMichalopoulos, Philip Robins, David Card.
Financial Work Incentive on Employment and Income(SRDC). 2000. Charles Michalopoulos, David Card,Lisa Gennetian, Kristen Harknett, Philip K. Robins.
The Self-Sufficiency Project at 36 Months: Effects onChildren of a Program That Increased ParentalEmployment and Income (SRDC). 2000. PamelaMorris, Charles Michalopoulos.
Mandatory Welfare Employment ProgramsNational Evaluation of Welfare-to-WorkStrategiesConceived and sponsored by the U.S. Department ofHealth and Human Services, with support from the U.S.Department of Education, this is the largest-scaleevaluation ever conducted of different strategies formoving people from welfare to employment.
Adult Education for People on AFDC: A Synthesis ofResearch (U.S. Department of Education [ED]/U.S.Department of Health and Human Services [HHS]).1995. Edward Pauly.
Early Findings on Program Impacts in Three Sites(HHS/ED). 1995. Stephen Freedman, DanielFriedlander.
Five Years After: The Long-Term Effects of Welfare-to-Work Programs (Russell Sage Foundation). 1995.Daniel Friedlander, Gary Burtless.
Monthly Participation Rates in Three Sites and FactorsAffecting Participation Levels in Welfare-to-WorkPrograms (HHS/ED). 1995. Gayle Hamilton.
Changing to a Work First Strategy: Lessons from LosAngeles County's GAIN Program for WelfareRecipients. 1997. Evan Weissman.
Evaluating Two Welfare-to-Work Program Approaches:Two-Year Findings on the Labor Force Attachment andHuman Capital Development Programs in Three Sites(HHS/ED). 1997. Gayle Hamilton, Thomas Brock,Mary Farrell, Daniel Friedlander, Kristen Harknett.
Work First: How to Implement an Employment-FocusedApproach to Welfare Reform. 1997. Amy Brown.
Implementation, Participation Patterns, Costs, andTwo-Year Impacts of the Portland (Oregon) Welfare-to-Work Program (HHS/ED). 1998. Susan Scrivener,Gayle Hamilton, Mary Farrell, Stephen Freedman,Daniel Friedlander, Marisa Mitchell, Jodi Nudelman,Christine Schwartz.
Do Mandatory Welfare-to-Work Programs Affect theWell-Being of Children? A Synthesis of ChildResearch Conducted as Part of the NationalEvaluation of Welfare-to-Work Strategies (HHS/ED).2000. Gayle Hamilton.
Evaluating Alternative Welfare-to-Work Approaches:Two-Year Impacts for Eleven Programs (HHS/ED).2000. Stephen Freedman, Daniel Friedlander, GayleHamilton, JoAnn Rock, Marisa Mitchell, JodiNudelman, Amanda Schweder, Laura Storto.
Impacts on Young Children and Their Families TwoYears After Enrollment: Findings from the ChildOutcomes Study (HHS/ED). 2000. Sharon McGroder,Martha Zaslow, Kristin Moore, Suzanne LeMenestrel.
What Works Best for Whom: Impacts of 20 Welfare-to-Work Programs by Subgroup (HHS/ED). 2000.Charles Michalopoulos, Christine Schwartz.
Los Angeles's Jobs-First GAIN ProgramAn evaluation of Los Angeles's refocused GAIN(welfare-to-work) program, which emphasizes rapidemployment. This is the first in-depth study of a full-scale "work first" program in one of the nation's largesturban areas.
Changing to a Work First Strategy: Lessons from LosAngeles County's GAIN Program for WelfareRecipients. 1997. Evan Weissman.
The Los Angeles Jobs-First GAIN Evaluation:Preliminary Findings on Participation Patterns andFirst-Year Impacts. 1998. Stephen Freedman, MarisaMitchell, David Navarro.
The Los Angeles Jobs-First GAIN Evaluation: First-Year Findings on Participation Patterns and Impacts.1999. Stephen Freedman, Marisa Mitchell, DavidNavarro.
The Los Angeles Jobs-First GAIN Evaluation: FinalReport on a Work First Program in a Major UrbanCenter. 2000. Stephen Freedman, Jean Knab, LisaGennetian, David Navarro.
-97- 121
Teen Parents on WelfareTeenage Parent Programs: A Synthesis of the Long-
Term Effects of the New Chance Demonstration,Ohio 's Learning, Earning, and Parenting (LEAP)Program, and the Teenage Parent Demonstration(TPD). 1998. Robert Granger, Rachel Cytron.
Ohio's LEAP ProgramAn evaluation of Ohio's Learning, Earning, and
Parenting (LEAP) Program, which uses financialincentives to encourage teenage parents on welfare tostay in or return to school.
LEAP: Final Report on Ohio 's Welfare Initiative toImprove School Attendance Among Teenage Parents.1997. Johannes Bos, Veronica Fellerath.
New Chance DemonstrationA test of a comprehensive program of services that seeks
to improve the economic status and general well-beingof a group of highly disadvantaged young women andtheir children.
New Chance: Final Report on a Comprehensive Programfor Young Mothers in Poverty and Their Children.1997. Janet Quint, Johannes Bos, Denise Polit.
Parenting Behavior in a Sample of Young Mothers inPoverty: Results of the New Chance ObservationalStudy. 1998. Martha Zaslow, Carolyn Eldred, editors.
Focusing on Fathers
Parents' Fair Share DemonstrationA demonstration for unemployed noncustodial parents
(usually fathers) of children on welfare. PFS aims toimprove the men's employment and earnings, reducechild poverty by increasing child support payments,and assist the fathers in playing a broader constructiverole in their children's lives.
Low-Income Parents and the Parents' Fair ShareDemonstration. 1996. Earl Johnson, Fred Doolittle.
Working with Low-Income Cases: Lessons for the ChildSupport Enforcement System from Parents FairShare. 1998. Fred Doolittle, Suzanne Lynn.
Building Opportunities, Enforcing Obligations:Implementation and Interim Impacts of Parents ' FairShare. 1998. Fred Doolittle, Virginia Knox, CynthiaMiller, Sharon Rowser.
Fathers Fair Share: Helping Poor Men Manage ChildSupport and Fatherhood (Russell Sage Foundation).1999. Earl Johnson, Ann Levine, Fred Doolittle.
Parenting and Providing: The Impact of Parents FairShare on Paternal Involvement. 2000. Virginia Knox,Cindy Redcross.
Working and Earning: The Impact of Parents FairShare on Low-Income Fathers' Employment. 2000.John M. Martinez, Cynthia Miller.
The Responsible Fatherhood Curriculum. 2000. EileenHayes, with Kay Sherwood.
OtherCan They All .Work? A Study of the Employment
Potential of Welfare Recipients in a Welfare-to-WorkProgram. 1995. James Riccio, Stephen Freedman.
Florida's Project Independence: Benefits, Costs, and Two-Year Impacts of Florida's JOBS Program. 1995. JamesKemple, Daniel Friedlander, Veronica Fellerath.
From Welfare to Work Among Lone Parents in Britain:Lessons for America. 1996. James Riccio.
Education ReformCareer AcademiesThe largest and most comprehensive evaluation of aschool-to-work initiative, this study examines apromising approach to high school restructuring and theschool-to-work transition.
Career Academies: Early Implementation Lessons from a10-Site Evaluation. 1996. James Kemple, JoAnn LeahRock.
Career Academies: Communities of Support for Studentsand Teachers Emerging Findings from a 10-SiteEvaluation. 1997. James Kemple.
Career Academies: Building Career Awareness andWork-Based Learning Activities Through EmployerPartnerships. 1999. James Kemple, Susan Poglinco,Jason Snipes.
Career Academies: Impacts on Students' Engagementand Performance in High School. 2000. JamesKemple, Jason Snipes.
Project GRADThis evaluation examines Project GRAD, an educationinitiative targeted at urban schools and combining anumber of proven or promising reforms.Building the Foundation for Improved Student
Performance: The Pre-Curricular Phase of ProjectGRAD Newark. 2000. Sandra Ham, Fred C. Doolittle,Glee Ivory Holton.
LILAA InitiativeThis study of the Literacy in Libraries Across America(LILAA) initiative explores the efforts of five adultliteracy programs in public libraries to improve learnerpersistence.So I Made Up My Mind: Introducing a Study of Adult
Learner Persistence in Library Literacy Programs.2000. John T. Comings, Sondra Cuban.
Project TransitionA demonstration program that tested a combination ofschool-based strategies to facilitate students' transitionfrom middle school to high school.
Project Transition: Testing an Intervention to HelpHigh School Freshmen Succeed. 1999. Janet Quint,Cynthia Miller, Jennifer Pastor, Rachel Cytron.
-98-re.
(-1, 4
Equity 2000Equity 2000 is a nationwide initiative sponsored by theCollege Board to improve low-income students' accessto college. The MDRC paper examines theimplementation of Equity 2000 in Milwaukee PublicSchools.
Getting to the Right Algebra: The Equity 2000 Initiativein Milwaukee Public Schools. 1999. Sandra Ham,Erica Walker.
School-to-Work ProjectA study of innovative programs that help students makethe transition from school to work or careers.
Home-Grown Lessons: Innovative Programs LinkingSchool and Work (Jossey-Bass Publishers). 1995.Edward Pauly, Hilary Kopp, Joshua Haimson.
Home-Grown Progress: The Evolution of InnovativeSchool-to-Work Programs. 1997. Rachel Pedraza,Edward Pauly, Hilary Kopp.
Employment and CommunityInitiativesConnections to Work ProjectA study of local efforts to increase competition in thechoice of providers of employment services for welfarerecipients and other low-income populations. Theproject also provides assistance to cutting-edge localinitiatives aimed at helping such people access andsecure jobs.
Tulsa's IndEx Program: A Business-Led Initiative forWelfare Reform and Economic Development. 1997.Maria Buck.
Washington Works: Sustaining a Vision of WelfareReform Based on Personal Change, WorkPreparation, and Employer Involvement. 1998. SusanGooden.
Cost Analysis Step by Step: A How-to Guide forPlanners and Providers of Welfare-to-Work andOther Employment and Training Programs. 1998.David Greenberg, Ute Appenzeller.
Designing and Administering a Wage-PayingCommunity Service Employment Program UnderTANF: Some Considerations and Choices. 1999. KaySherwood.
San Francisco Works: Toward an Employer-LedApproach to Welfare Reform and WorkforceDevelopment. 2000. Steven Bliss.
Jobs-Plus InitiativeA multi-site effort to greatly increase employment
among public housing residents.
A Research Framework for Evaluating Jobs-Plus, aSaturation and Place-Based Employment Initiativefor Public Housing Residents. 1998. James Riccio.
Mobilizing Public Housing Communities for Work:Origins and Early Accomplishments of the Jobs-PlusDemonstration. 1999. James Riccio.
Building a Convincing Test of a Public HousingEmployment Program Using Non-ExperimentalMethods: Planning for the Jobs-Plus Demonstration.1999. Howard Bloom.
Jobs-Plus Site-by-Site: An Early Look at ProgramImplementation. 2000. Edited by Susan PhilipsonBloom with Susan Blank.
Section 3 Public Housing StudyAn examination of the effectiveness of Section 3 of the1968 Housing and Urban Development Act in affordingemployment opportunities for public housing residents.
Lessons from the Field on the Implementation of Section 3(U.S. Department of Housing and Urban Development).1996. Maxine Bailey, Suzanne Lynn.
Canada's Earnings Supplement ProjectA test of an innovative financial incentive intended toexpedite the reemployment of displaced workers andencourage full-year work by seasonal or part-yearworkers, thereby also reducing receipt ofUnemployment Insurance.
Implementing the Earnings Supplement Project: A Testof a Re-employment Incentive (Social Research andDemonstration Corporation). 1997. Howard Bloom,Barbara Fink, Susanna Lui-Gurr, Wendy Bancroft,Doug Tattrie.
Testing a Re-employment Incentive for DisplacedWorkers: The Earnings Supplement Project. 1999.Howard Bloom, Saul Schwartz, Susanna Lui-Gurr,Suk-Won Lee.
MDRC Working Papers on ResearchMethodologyA new series of papers that explore alternative methodsof examining the implementation and impacts ofprograms and policies.
Building a Convincing Test of a Public HousingEmployment Program Using Non-ExperimentalMethods: Planning for the Jobs-Plus Demonstration.1999. Howard Bloom.
Estimating Program Impacts on Student AchievementUsing "Short" Interrupted Time Series. 1999.Howard Bloom.
Using Cluster Random Assignment to Measure ProgramImpacts: Statistical Implications for the Evaluation ofEducation Programs. 1999. Howard Bloom, JohannesBos, Suk-Won Lee.
-99-
123
About MDRC
The Manpower Demonstration Research Corporation (MDRC) is a nonprofit,nonpartisan social policy research organization. We are dedicated to learning whatworks to improve the well-being of low-income people. Through our research andthe active communication of our findings, we seek to enhance the effectiveness ofsocial policies and programs. MDRC was founded in 1974 and is located in NewYork City and San Francisco.
MDRC's current projects focus on welfare and economic security, education, andemployment and community initiatives. Complementing our evaluations of a widerange of welfare reforms are new studies of supports for the working poor andemerging analyses of how programs affect children's development and theirfamilies' well-being. In the field of education, we are testing reforms aimed atimproving the performance of public schools, especially in urban areas. Finally, ourcommunity projects are using innovative approaches to increase employment inlow-income neighborhoods.
Our projects are a mix of demonstrations field tests of promising programmodels and evaluations of government and community initiatives, and weemploy a wide range of methods such as large-scale studies to determine aprogram's effects, surveys, case studies, and ethnographies of individuals andfamilies. We share the findings and lessons from our work including bestpractices for program operators with a broad audience within the policy andpractitioner community, as well as the general public and the media.
Over the past quarter century, MDRC has worked in almost every state, all of thenation's largest cities, and Canada. We conduct our projects in partnership with stateand local governments, the federal government, public school systems, communityorganizations, and numerous private philanthropies.
124
aheTlplitext .eration
MDRC
16 East 34 StreetNew York, New York 10016
(212) 532-3200
www.mdrc.org
88 Kearny Street, Suite 1800San Francisco, California 94108
(415) 781-3800
U.S. Deportment of EducofionOffice of Educalioncrl Research and Improvement (081)
National Library of Education (NL.Educational Resources Informatibn Center (81C)
NOTICE
REPRODUCTION BASIS
ERI
This document is covered by a signed "Reproduction Release(Blanket) form (on file within the ERIC system), encompassing allor classes of documents from its source organization and, therefore,does not require a "Specific Document" Release form.
This document is Federally-fimded, or carries its own permission toreproduce, or is otherwise in the public domain and, therefore, maybe reproduced by ERIC without a signed Reproduction Release farm(either "Specific Document" or "Blanket").,
EFL -089 (9/97)
top related